Gynecologic Cancer

Abstract

The management of gynecologic cancers has historically been guided by a clinically-oriented staging system, based largely on physical examination and standard imaging studies including CT of the abdomen and pelvis. This has more recently been supplemented by pre-treatment MRI and functional imaging, as well as imaging biomarkers. This chapter will focus on the most common gynecologic malignancies, discussing the clinical, pathological, and treatment-related factors that influence clinical outcome as well as the influence of biomarkers on prognosis.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Gynecologic cancers are a diverse group of tumors which are characterized by an orderly pattern of loco-regional spread that is correlated with prognosis. This is reflected in the International Federation of Gynecology and Obstetrics (FIGO) staging system. In addition to the traditional staging system, biologic and molecular markers reflecting angiogenesis, hypoxia and tumor cell proliferation are emerging that correspond to treatment response and prognosis. However, their use in clinical decision-making remains limited. Imaging-based predictors are easier to utilize in clinical management, and show promise in predicting outcome and risk of failure both before and during therapy.

At present, in patients treated primarily with surgery, histopathologic factors can be highly predictive of treatment outcomes, and these variables can dictate the need for adjuvant radiation therapy and/or chemotherapy. For example, histopathologic variables such as margin status, lymphovascular invasion, and positive lymph nodes are commonly factored into adjuvant therapy decisions (Sedlis et al. 1999; Peters et al. 2000). In the future, there may be a role for biomarkers in this regard. However, to date, there has not been substantial progress in the use of biomarkers for use in the post-operative setting.

2 Cancer of the Cervix

The prognosis and treatment outcome for patients with cervical cancer are largely determined by local tumor extent, tumor size and regional lymphatic spread, which follows predictable pathways along anatomic routes and lymph node echelons. In general, the extent of loco-regional spread will guide the selection of therapy—tumors confined to the cervix are managed primarily by surgical therapy, while those with extension to the parametrium, distal vagina or adjacent organs are treated by primary radiation and chemotherapy.

2.1 Staging

The loco-regional tumor extent is only partially reflected by the traditional FIGO staging system (Pecorelli 2009). The FIGO staging system (Table 1) relies on findings from clinical examination and invasive investigations, including cystoscopy and proctoscopy, with biopsy. It also allows the use of radiographic information from plain X-ray films. However, information regarding lymph node involvement, a strong determinant of outcome, is not incorporated. In addition, findings from CT, MRI and functional imaging, namely PET-CT, are not utilized. This results in inherent limitations in assessing well validated prognostic factors, including tumor volume, involvement of adjacent structures, and parametrial extension, that are challenging to assess by palpation and visual inspection alone. In addition, the detection of regional lymph node spread and sites of distant metastatic disease, that can be detected with cross-sectional imaging, can be missed (Eifel 1994). Due to these inherent limitations, FIGO staging has been shown to result in under-staging of 20–60 % of cervical cancer patients, when compared with surgical staging (Averette et al. 1975). This explains why significant variations in treatment failure rates and survival are observed within each FIGO stage category (Eifel et al. 1994; Perez et al. 1992a).

Table 1 FIGO and TNM staging of cervical cancer

Despite the fact that the FIGO staging system suffers from the aforementioned limitations, it remains the current standard of practice, and provides the major entrance criteria utilized in determining the eligibility of patients for cooperative group trials. Thus, most current cooperative group trials enroll patients across almost the entire FIGO stage spectrum, from stage IB2-IVA, and accession them to largely uniform treatment regimens. Of note, although cross-sectional imaging is not “permitted” to influence FIGO stage assignment, the use of CT, MRI and molecular imaging with fluorodeoxyglucose (¹⁸F)FDG PET imaging is likely to result in “stage migration” by excluding patients with subtle imaging-based evidence of regional or distant metastatic involvement from cooperative group trials (Fig. 1). This will make any improvements of therapy with newer interventions difficult to compare to historic controls. However, the incorporation of functional imaging into future clinical trials will potentially enhance our ability to accurately stratify patients and tailor therapy to the “true” clinical stage (i.e. locally advanced vs. metastatic disease).

Fig. 1

PET-CT and Staging of Cervical Cancer: A 43 year old woman with invasive squamous cell carcinoma of the cervix underwent PET-CT revealing retroperitoneal and supraclavicular adenopathy, consistent with Stage IV disease

2.2 Clinical Factors

Eligibility for primary surgical therapy is determined by regional tumor extent to adjacent structures and significantly influences prognosis. Patients with stage I disease (tumor limited to the cervix) and selected patients with stage II disease (including patients with upper vaginal involvement), are candidates for radical hysterectomy. The overall survival of surgically treated patients with stage IB tumors ranges from 85 to 90 % (Morley and Seski 1976; Hopkins and Morley 1991; Landoni et al. 1997). However, large tumor size, deep cervical invasion, lymphovascular space invasion, as well as involved lymph nodes and parametrial involvement have been recognized as risk factors for pelvic recurrence after radical hysterectomy. Depending on the number and extent of these factors present, adjuvant therapy can improve outcomes, albeit at the cost of increased risk of toxicity from adjuvant radiation and/or chemotherapy (Sedlis et al. 1999; Peters et al. 2000; Rotman et al. 2006). Thus, if imaging modalities or other factors could identify the presence of these pathologic features during workup leading to upstaging, definitive radiation and chemotherapy could be considered instead of primary surgery, thus potentially reducing the morbidity of treatment.

2.2.1 Stage

In patients with cervical cancer treated with definitive radiation therapy, FIGO stage remains an important prognostic factor. Due to the relative rarity of cervical cancer in western nations, phase III cooperative group trials do not subclassify patients by stage, nor do they group patient cohorts as stages IB–II versus III–IVA for subgroup analyses, because analysis by individual stage categories would require unachievably large patient cohorts. Based on large single-institution series in which contemporary radiation techniques and concurrent chemotherapy were utilized, reported local control rates, disease free survival rates, and overall survival rates for patients with Stage IB–IIA and III–IVA are 87 and 79 %, 74 and 54 %, and 79 and 59 %, respectively (Whitney et al. 1999; Eifel et al. 2004; Rose et al. 2007).

2.2.2 Tumor Volume

In addition to FIGO stage, tumor size has profound prognostic significance (Eifel et al. 1994; Kovalic et al. 1991). In 1988 FIGO added tumor diameter as a stratifying factor for stage I disease, with tumors less than or greater than 4 cm classified as stage IB1 versus IB2 respectively. In the 2009 revision of the staging system, tumor size of greater or less than 4 cm was also incorporated into the stage IIB category (Pecorelli et al. 2009). Tables 2 and 3 shows the profound significance of tumor size, measured as largest or average palpated diameter, for local control and survival. Within the same stage category of IB, tumor size of <4 cm in diameter was associated with a disease-free survival of 87 %, compared to 72 % for 6 cm, 69 % for 7 cm, 64 % for 8 cm and 47 % for >8 cm tumors (Table 2). Similar relationship exists for tumor size and outcomes within the stage IIB category (Table 3) (Eifel et al. 1994; Hansgen and Dunst 1996; Hockel et al. 1996; Homesley et al. 1980; Lowrey et al. 1992; Perez et al. 1992b; Mendenhall et al. 1984).

Table 2 Tumor diameter versus disease free survival for stage IB cervical cancer

Table 3 Tumor diameter versus disease free survival for stage IIB cervical cancer

2.2.3 Lymph Node Status

For any given FIGO stage, lymph node involvement reduces overall survival by approximately 50 % (Stehman et al. 1991). Furthermore, among patients with positive lymph nodes, prognosis declines with increasing extent of lymph node involvement (Macdonald et al. 2009; Hsu et al. 1972; Tsai et al. 1999; Takeda et al. 2002; Morice et al. 1999). In a pooled study by the Gynecologic Oncology Group (GOG), para-aortic involvement was associated with an 11-fold risk of recurrence and sixfold risk of death, and was also associated with extrapelvic failures (Berman et al. 1984). However, even with para-aortic lymph node involvement, survival in the range of 20–50 % has been reported for patients with locally advanced disease (Komaki et al. 1983; Rotman et al. 1994), justifying aggressive therapy for patients with regional lymphatic spread.

Although controversy exists whether surgical excision of suspicious lymph nodes improves outcomes, a large retrospective study of patients treated in the pre-chemo-radiation era showed among patients who underwent lymphadenectomy and postoperative radiation, patients with macroscopically involved lymph nodes had similar regional and distant tumor control as those with microscopic lymph node involvement, and significantly better than those patients with unresectable lymph nodes (Cosin et al. 1998). This supports the use of imaging for identification of involved nodes, thus allowing for a tumor directed combined modality approach. Increasing use of molecular imaging in cervix cancer will facilitate this approach and will also likely lead to stage migration as lymph nodes with more subtle involvement can be identified and treated more aggressively.

2.3 Patient Factors

2.3.1 Hemoglobin

Over the past 50 years, numerous studies have provided indirect evidence that the effects of poor tumor blood supply have an adverse impact on radiation response. Early studies of morphologic parameters of angiogenesis, such as microvessel density, have been shown to correlate with radio-responsiveness and clinical outcome in cervical cancer (Awwad et al. 1986; Cooper et al. 1998). Cervical cancer patients with high inter-capillary distances locally within their tumors measured by colposcopy were found to have increased tumor recurrence rates after radiation therapy (Kolstad 1968).

Similarly cervical cancer patients with low hemoglobin levels have been reported to have higher recurrence rates after radiotherapy (Mendenhall et al. 1984; Bush et al. 1978; Evans and Bergsjo 1965; Diesche et al. 1983; Thomas 2001; Dunst et al. 2003). This supports the concept that poor “systemic” oxygenation is clinically significant for treatment outcome. Haensgen et al. analyzed hemoglobin levels of 70 patients, and reported survival was 27 % for patients with low hemoglobin (<11 g/dL), compared to 62 % in those with higher levels (Haensgen et al. 2001). Dunst et al. mirrored these results, showing overall survival of 64 and 32 %, respectively and local recurrence rates of 15 % versus 67 %, respectively (Dunst et al. 2003). Hemoglobin during the course of therapy, when the actual cytotoxic events occur, may also be relevant. Thomas et al. showed in 605 patients that the average weekly hemoglobin nadir <12 g/dL was associated with a higher incidence of local failure and metastases (Thomas 2001). In a recent study, weekly mean hemoglobin levels measured during the course of radiotherapy was more predictive of outcome than pre-therapy or nadir hemoglobin (Mayr et al. 2009). In all, the thresholds value for this effect of hemoglobin level appears to be in the range of 11–12 g/dL.

Although the impact of blood transfusion on outcome in patients treated with definitive radiation therapy remains controversial, the Canadian experience suggests that maintaining hemoglobin levels above 12 g/dL is associated with improved 5-year survival. Pre-treatment hemoglobin, which may not reflect the longitudinal status of hemoglobin levels, did not have any impact on outcome (Grogan et al. 1999). Interestingly, in one retrospective study of 204 patients at a single institution where departmental practice was to transfuse for hemoglobin <11 g/dL, it was noted that only 18.5 % of patients who received transfusion had a sustained response to transfusion, although outcomes for these patients were equivalent to those presenting with normal hemoglobin (Kapp et al. 2002). However, for patients who did not have a sustained response to blood transfusion, outcomes were significantly worse compared to those with response or with normal hemoglobin pre-therapy. While there was a therapeutic benefit to transfusion for those patients with sustained response, the low rate of response of 18.5 % was disappointing, and it was proposed that finding and treating the underlying cause of the anemia may be more beneficial.

2.4 Histologic Factors

2.4.1 Histology

Approximately 90 % of cervical cancers are squamous cell carcinomas. Squamous cell carcinomas arise from epithelial precursors, and can be classified into one of three cell types: large cell keratinizing, large cell nonkeratinizing, and small cell. Tumor grade is based on the degree of differentiation, and is reported as well, moderately, or poorly differentiated. Adenocarcinoma is the second most common, accounting for 10–15 % depending on region and age. More recently, the incidence of adenocarcinomas appears to be increasing, especially in younger patients (Liu et al. 2001; Smith et al. 2000). Adenocarcinomas arise from the mucus-secreting endocervical glands of the cervix or the cylindrical mucosa. The most common subtype of adenocarcinoma of the cervix is endometrioid adenocarcinoma, where cells have characteristic features of the endometrium and grading is based on the degree of gland formation. It is critical to differentiate this from primary endometrioid endometrial adenocarcinoma as recommended therapy would change, thus clinical presentation, such as absence or presence of an endometrial tumor with extension into the cervix, is incorporated to determine the true site of primary disease. The next most common subtype of adenocarcinoma is adenosquamous histology, comprising 21–30 % of adenocarcinomas (Farley et al. 2003; Kleine et al. 1989), and is characterized by epithelial cell cores mixed with glandular structures. Other histologies, such as clear cell, small cell carcinoma, basaloid carcinoma, lymphoma, and sarcomas occur, but are rare and have varying prognostic impact.

The prognosis of adenocarcinoma versus squamous cell histology is debated. While adenocarcinoma is associated with an increased risk of failure, particularly metastatic failure in some retrospective reports (Eifel et al. 1995; Huang et al. 2011, 2012), many show no significant impact on outcome between adenocarcinoma and squamous cell carcinoma (Shingleton et al. 1995; Look et al. 1996; Davidson et al. 1989). Interestingly, adenosquamous carcinoma may be associated with poorer recurrence free and overall survival (Farley et al. 2003; Look et al. 1996; Lea et al. 2003; Grisaru et al. 2001; Galic et al. 2012). In all, the differences in outcomes among these studies may be in part due to regional variation in Human Papilloma Virus (HPV) genotype distribution, changes in etiology and incidence of histologic type, differences in treatment approach, and overall study sizes, making it difficult to draw any definitive conclusion about subtype implications in the absence of prospective data.

2.4.2 Histopathologic Risk Factors in Postoperative Patients

In surgically treated stage I-IIA patients, lymph node involvement, parametrial invasion and involved margins have long been recognized as high risk factors for local recurrence and death (Morrow 1980). In those with involved lymph nodes, number of involved nodes (<3 vs. >3), bilaterality, level (common iliac vs. pelvic) and size (micro- vs. macroscopic) impact outcome (van Bommel et al. 1987; Tanaka et al. 1984). Therefore, adjuvant therapy based on histopathologic risk factors is paramount because salvage therapy for recurrent cervical cancer after hysterectomy has dismal results with a 5–45 % survival (Thomas et al. 1993). Postoperative radiation has been the hallmark in adjuvant therapy.

Tumor size, depth of invasion and capillary-lymphatic space invasion have also been shown to impact prognosis in surgically treated stage I-IIA patients. However, until the completion of the phase III GOG 92 study, the impact of adjuvant therapy on survival was not well established. GOG 92 (Sedlis et al. 1999; Rotman et al. 2006) established a set of intermediate risk factors (commonly referred to as “Sedlis criteria”) for poor outcome in stage IB patients treated with radical hysterectomy. Patients with two of the three features (capillary lymphatic space invasion, large clinical tumor diameter, or more than one-third cervical stromal invasion) were randomized to pelvic radiotherapy versus no further therapy (Table 4). At 10 years median follow-up, postoperative radiation reduced the risk of recurrence by 46 % (HR 0.54) with the greatest benefit in patients with a combination of deep 1/3 invasion plus tumor size >4 cm (HR 0.16) or capillary lymphatic space invasion plus deep 1/3 invasion with any tumor size (HR 0.53) (Rotman et al. 2006). There was no significant improvement in overall survival (Table 5). On subgroup analysis, proportionally greater improvement was noted among 44 patients with adenosquamous or adenocarcinoma, where adjuvant radiation therapy reduced the recurrence rate from 44 to 9 % (Rotman et al. 2006). A current GOG study is underway to evaluate whether postoperative radiation with concurrent chemotherapy can further improve upon this outcome.

Table 4 Inclusion criteria for GOG 92: randomization to postoperative pelvic radiotherapy versus no further therapy in stage IB intermediate risk cervical cancer

Table 5 GOG 92 results: postoperative radiotherapy improves recurrence-free, but not overall survival, in intermediate risk stage IB cervix cancer

Improvement of adjuvant therapy with the addition of concurrent chemotherapy to radiation has also been demonstrated for some select patients. The intergroup trial GOG 109 randomized stage IA2-IIA patients treated with radical hysterectomy and pelvic lymphadenectomy and high risk features, defined as positive pelvic lymph nodes and/or positive margins, and/or microscopic involvement of the parametrium, to pelvic radiotherapy versus pelvic radiotherapy with chemotherapy (cisplatin/5-FU for 4 cycles during and after radiation) (Peters et al. 2000). Addition of chemotherapy resulted in significant improvement of overall survival at 81 % versus 71 % (HR 1.96, p = 0.007). The greatest benefit was observed for patients with larger tumors and multiple involved lymph nodes, underscoring the importance of identification of involved lymph nodes in order to offer the optimal adjuvant therapy. Table 6 summarizes the results of five randomized trials that show improved survival with concurrent chemotherapy and radiotherapy (Peters et al. 2000; Whitney et al. 1999; Rose et al. 2007; Keys et al. 1999; Morris et al. 1999).

Table 6 Estimates of the relative risk of death in five clinical trials of radiotherapy and concurrent chemotherapy

2.4.3 Molecular Tumor Markers

2.4.3.1 HPV

HPV is found in an estimated 93–99.7 % of invasive cervical cancer (Bosch et al. 1995; Walboomers et al. 1999). Further, the prevalence of different genotypes varies in cellular histology. HPV16 is identified in the majority of squamous cell carcinomas, and HPV18 is the predominant genotype in adenocarcinomas and adenosquamous carcinomas (Bosch et al. 1995). HPV may be a prognostic indicator for outcomes. Several studies have shown HPV18 and HPV16 is associated with more advanced cervical cancers at presentation and poorer outcomes (Schwartz et al. 2001; Pilch et al. 2001; Burger et al. 1996). Further, HPV18 has been associated with increased radioresistance and increased recurrence rates compared to other HPV genotypes in patients receiving only radiation therapy (Wang et al. 2010). However, HPV18 has subsequently been shown to be predictive of improved disease specific survival when concurrent chemotherapy and radiotherapy was used in place of radiotherapy alone (Wang et al. 2012). The clinical utility of this association is an area of active investigation.

2.4.3.2 Angiogenesis

Angiogenesis-related molecular markers would be expected to be of great importance for radiation and chemotherapy because of the critical dependence of the cytotoxic effect on tumor microcirculation and oxygenation (Tannock 1972). It is postulated that poorly-perfused, hypoxic, endophytic tumors are associated with radio-resistance and resulting poor treatment outcome in cervical cancer. Angiogenic factors have been shown to correlate with tumor recurrence and survival in surgically treated patients (Cheng et al. 2000; Dellas et al. 1997; Dinh et al. 1996; Hawighorst et al. 1997; Lee et al. 2011; Mayr et al. 1999; Kainz et al. 1995; Obermair et al. 1998; Tjalma et al. 2000). Therefore, there has been increasing interest in molecular markers of angiogenesis and cytokines in cervical cancer. Cooper et al. (Cooper et al. 1998) reported that patients with high MVD had significantly poorer local control and survival. Although Gaffney et al. (2003) found increased VEGF and EGFR expression to be associated with poor survival, inconsistent results have been observed in regard to VEGF association with tumor progression, stage (Loncaster et al. 2000), histologic type (Cheng et al. 2000; Loncaster et al. 2000, 2002) and microvessel density (MVD) (Mayr et al. 1999; Hawighorst et al. 1998). High expression of another angiogenic marker, carbonic anhydrase IX (CA IX) correlates with poor survival (Loncaster et al. 2002). More recently the GOG evaluated a panel of angiogenesis markers including MVD, VEGF, CD31 (non-specific endothelial marker), TSP-1 (thrombospondin-1 an anti-angiogenesis factor), and CD105 (tumor-specific endothelial marker) and association with clinical outcome (Randall et al. 2009). Expression of each was determined in tumors from patients included in GOG 109, including stage IA2-IIA patients with positive lymph nodes, parametrial involvement, or positive surgical margins (Peters et al. 2000). Of these, only high expression CD31 was independently predictive of improved disease free and overall survival. Authors posit that this may be representative of CD31 as a surrogate marker for improved tumor flow and oxygenation, thus improving response to adjuvant therapy.

2.4.3.3 Alternate Candidate Molecules

There has been increasing interest in evaluation of molecular mechanisms of radiation response through candidate gene approach and microarray analysis. Studies in cervical cancer cell lines have found that genes related to angiogenesis, apoptosis and tumor cell invasion correlate with radio-resistance (Harima et al. 2004; Kitahara et al. 2002; Tewari et al. 2005; Wong et al. 2003). A pilot study of 12 patients in 2008 used microarray analysis and demonstrated immortalization upregulated protein (IMUP), IGF-2, and ARHD were associated with tumor recurrence in patients treated with radiation and concurrent chemotherapy (Klopp et al. 2008). Proteins that have been shown to correlate with clinical outcome include Ku80, GADD45 (Harima et al. 2003), bax, bcl-2 (Harima et al. 1998), intracellular adhesion molecule-3 (ICAM-3) (Chung et al. 2005), and hypoxia inducible factor (HIF)-1a (Bachtiary et al. 2003; Burri et al. 2003). However, to date, none of these molecular markers has been incorporated into clinical care.

2.5 Imaging Prognostic/Predictive Markers

2.5.1 Morphologic Imaging

Improvements in spatial and temporal resolution of cross-sectional imaging have broadened the capabilities of both anatomical and functional imaging in cervical cancer. Three-dimensional tumor volume can be quantified, and tumor extent and involvement of adjacent structures more accurately assessed than by clinical palpation (Hricak et al. 1988; Hricak 1991; Bhosale et al. 2010; Balleyguier et al. 2011). Higher temporal and spatial resolution also allows for functional imaging, such as dynamic contrast-enhanced (DCE) MRI and diffusion-weighted imaging, in addition to the morphologic/anatomical imaging. Beyond pre-therapy assessment, repeated imaging throughout the course of definitive chemoradiotherapy with an intact cervix provides longitudinal information on functional changes in response to ongoing therapy. Such on-therapy imaging shows promise for deriving imaging biomarkers to predict therapeutic response and disease outcome.

Tumor size in cervical cancer is best assessed with MRI (Bhosale et al. 2010; Balleyguier et al. 2011), which was demonstrated in imaging-histologic correlation studies (Burghardt et al. 1989; Greco et al. 1989). For on-therapy assessments, the velocity of tumor regression, assessed by 3D tumor volumetry (not diameter-based measurement) allows an indirect measure of therapy responsiveness (Mayr et al. 2006), which has been shown to be predictive of treatment outcome in cervical cancer patients treated with radiation/chemotherapy (Hatano et al. 1999; Mayr et al. 1996, 2010; Sethi et al. 2005; Lim et al. 2008). Using 3D volumetric measurements, Mayr et al. found that patients with <20 % of residual tumor volume at 40–50 Gy delivered over 4–5 weeks had excellent local control and disease free survival of 90.5 and 88.4 %, compared to 23.1 and 45.4 % in patients with slower tumor regression (Mayr et al. 1996). Similarly, Hatano et al. (1999) found 100 % local control in patients with rapid tumor volume regression to less than 30 % of the original volume at 30 Gy over 3 weeks. Further, the velocity of tumor shrinkage directly correlates with patients’ risk for local failure and death of disease (Mayr et al. 2010). Such early predictive information, available during the ongoing therapy course, may open a window of opportunity to adapt and intensify therapy. For post-therapy assessment in the early follow-up period, complete resolution of the tumor 3–6 months after therapy is associated with better outcome (Hricak 1991; Flueckiger et al. 1992).

2.5.2 Functional Imaging

Among the functional imaging modalities, DCE MRI provides an in vivo imaging biomarker that indirectly reflects tumor perfusion and the delivery of oxygen and therapeutic agents to the tumor. Low perfusion, indicative of poor vascularity and oxygenation, before or early during the course of radiation therapy (at approximately 20 Gy, ~2 weeks), significantly predicts unfavorable local tumor control (73 % vs. 100 %, p = 0.006) and survival (47 % vs. 79 %, p = 0.001, respectively). The 2-week intra-treatment time point may be superior to the pre-therapy time point likely because the 2-week DCE MRI incorporates early therapy-specific information of responsiveness to the ongoing treatment (Yuh et al. 2009).

Diffusion-weighed imaging, which indirectly assesses tumor cellularity (Hamstra et al. 2008; Ross et al. 2003) provides another imaging biomarker in cervix cancer. The apparent diffusion coefficient (ADC) measures the magnitude of diffusion (of water molecules) within tissues. A low ADC value is indicative of increased tissue cellularity, and an increase in the ADC suggests cell death. Such an ADC increase can occur very early, within days of therapy start, prior to any morphologic changes (e.g. tumor volume) (Charles-Edwards and DeSouza 2006; Charles-Edwards et al. 2008; Chenevert et al. 2000). Early clinical experience shows that increase in ADC during ongoing radiation and chemotherapy correlates with improved tumor response (Harry et al. 2008; Naganawa et al. 2005; Liu et al. 2009). These studies suggest that both DCE-MRI and DW-MRI may have value as early imaging biomarkers of radioresponsiveness in cervical cancer.

In addition to being the most accurate assessment of lymph node involvement, FDG-PET/CT has also been used to assess the primary tumor during/after therapy. Persistent metabolic activity of the tumor 3 months after therapy has been correlated with poor outcome (Kidd et al. 2007). However, the optimal imaging timing for FDG-PET is a subject of active investigation.

3 Cancer of the Uterine Corpus

Endometrial cancer is the most common gynecologic malignancy in the United States. In 2013, 49,500 cases of endometrial cancer are expected, accounting for approximately 6 % of female malignancies, with approximately 8,200 deaths anticipated, accounting for 3 % of all female cancer deaths (Siegel et al. 2013). Mean age at diagnosis in the United States is approximately 62 years old, consistent with a disease largely occurring in postmenopausal women. SEER data show approximately 70 % of cases are diagnosed as localized disease, with an 81.5 % 5-year survival for all stages, and 95.3 % for localized disease (Howlader et al. 2013). Risk factors for endometrial cancer include diabetes, obesity, hyperestrogenic state, nulliparity, tamoxifen use, early menarche or late menopause, and anovulatory cycles (Brinton et al. 1992). Certain genetic diseases, such as hereditary nonpolyposis colon cancer (HNPCC) or Lynch syndrome and Cowden disease, are associated with increased risk for endometrial cancer, with lifetime risks ranging from 10 to 60 % depending on disease and specific genetic mutation (Aarnio et al. 1995; Gustafson et al. 2007).

Most endometrial cancers are diagnosed during the workup of abnormal, or postmenopausal, vaginal bleeding. Pathologic diagnosis is essential, as both FIGO stage and FIGO histologic grade are prognostic for outcome and determine treatment. Thus, diagnosis is often made via endometrial biopsy or dilation and curettage for those patients in which endometrial biopsy is not possible or non-diagnostic. Endometrial cancers often arise within the endometrial layer, and spread by invasion into the myometrium. In more advanced disease, tumor can spread to the uterine serosa, adnexa, endocervical canal, peritoneal cavity, bowel, bladder, and other adjacent structures. Lymphatic drainage is to the pelvic lymph nodes (including the internal/external iliacs, common iliacs, obturator, presacral and parametrial), with direct spread to the para-aortic lymph nodes possible. FIGO staging requires surgical staging based on the at-risk areas of spread, therefore total hysterectomy and bilateral salpingo-oopherectomy, with or without lymph node dissection, is performed in most patients. Adjuvant therapy is then based on pathologic information that determines the stage and grade of each endometrial cancer, both of which are prognostic for patient outcome.

3.1 Staging

The gold standard for staging in endometrial cancer remains surgical staging as defined by FIGO (Creasman 2009), Table 7. Prior to the 1988 FIGO staging system, staging was clinical evaluation for tumor size, extent of disease (confined to uterus or pelvic extension), and bowel or bladder involvement. However, this was found to understage patients approximately 23 % of the time (Creasman et al. 1987). Therefore, FIGO staging was changed to incorporate surgical evaluation and subsequent pathologic information for staging which improved the prognostic accuracy of staging. Initially, myometrial invasion, cervical invasion (including endocervical glandular involvement), adnexal involvement, serosal involvement, positive peritoneal cytology, and lymph node status were factored into staging. On the last revision of the FIGO surgical staging for endometrial cancer (2009), peritoneal cytology and isolated endocervical glandular involvement have been removed from the criteria. Further, myometrial invasion, previously stratified into three levels of involvement, is now subdivided into only two categories; invasion of less than one-half or invasion of one-half or more of the myometrium. Lymph node positive disease is substratified to pelvic lymph node only, or para-aortic lymph node disease (IIIC1 vs. IIIC2). In summary, under 2009 FIGO staging, stage I disease now includes endometrial/myometrial only disease; stage II disease invades cervical stroma; stage III disease is a heterogenous group with IIIA including uterine serosa or adnexal involvement, IIIB involving the vagina, and IIIC1 versus IIIC2 denoting pelvic lymph node only versus any para-aortic lymph node positive disease; stage IV represents metastatic disease to other sites not included above.

Table 7 FIGO and American Joint Committee on Cancer (AJCC 7th edition) TNM staging for endometrial cancer

Surgical staging at minimum is to include total hysterectomy and bilateral salpingo-oopherectomy (BSO). The role of extended surgical staging, with sampling and/or dissection of the pelvic and para-aortic lymph nodes, is still debated. Given the significant prognostic importance of lymph node metastasis, many advocate for lymph node histologic evaluation, and some have suggested a possible therapeutic benefit to lymphadenectomy, although not been proven in a prospective manner. While older techniques for extended surgical staging required laparotomy, more modern techniques with laparoscopic assisted methods have yielded equivalent nodal yields with reduced morbidity for many experienced gynecologic oncologists (Eltabbakh 2002; Scribner et al. 2002). Given the fact that many women with endometrial cancer are elderly, obese, and have co-morbidities such as diabetes, hypertension, and coronary artery disease, concerns exist for increased risks of deep venous thrombosis, vascular injury, or pulmonary emboli in the postoperative setting. Further, extended surgical staging followed by adjuvant radiation therapy is reported by some to carry higher enteric morbidity than hysterectomy and radiation alone (Lewandowski et al. 1990). Thus, some point to the experience of PORTEC and ASTEC trials as data to support omission of routine lymphadenectomy in low and intermediate risk patients without clinical/palpable adenopathy. PORTEC-1 included intermediate risk stage I patients, all undergoing total hysterectomy and BSO without lymphadenectomy randomized to adjuvant radiotherapy versus observation with 80–85 % overall survival at 5 years (Creutzberg et al. 2000). In ASTEC, intermediate risk patients underwent total hysterectomy-BSO, pelvic washings, and para-aortic lymph node palpation and were randomized to lymphadenectomy or no further surgery, with no statistically significant difference on overall survival at 3 years (ASTEC study group et al. 2009). Conversely, several studies support the role of maximal surgical debulking and resection of gross nodal disease, with improvement in median survival in some cohorts from 8.8 to 37.5 months (Bristow et al. 2003; Chi et al. 1997; Lambrou et al. 2004).

Of note, the American College of Obstetricians Gynecologists (ACOG) recommends comprehensive surgical staging including total hysterectomy and BSO, pelvic washings, bilateral pelvic and paraaortic lymphadenectomy, and complete resection of all disease, with exceptions considered for young or perimenopausal women with grade 1 endometrioid adenocarcinoma associated with atypical endometrial hyperplasia and those at increased risk of morbidity/mortality secondary to comorbidities (American College of Obstetricians and Gynecologists 2005). Omental sampling is also often performed, especially in papillary serous and clear cell histology due to the risk of upper abdominal spread.

3.2 Clinical Factors

3.2.1 Stage

Surgical stage continues to be one the most important clinical factors predictive of outcomes. The outcomes of 81,900 patients with endometrial cancer from 1988 to 2006 in a SEER database and a cohort of 1,268 patients from the MoMaTEC study were shown to verify the improved prognostic utility of the current 2009 FIGO staging in comparison to the FIGO 1988 staging schema (Lewin et al. 2010; Werner et al. 2012). Five year overall survival rates in early stage disease were 90–96 %, 78–87 %, and 74–80 %, respectively, for stage IA and IB and stage II. In locally advanced disease, 5-year overall survival was 48–56 %, 36–53 %, 57–60 %, and 49–53 % for stage IIIA (serosa/adnexa), IIIB (vaginal), IIIC1 (pelvic lymph node), and IIIC2 (para-aortic lymph node), respectively. Survival in stage IV disease ranged from 16 to 57 %.

3.2.2 Lymph Node Status

Lymph node status is incorporated in the staging classification above. A drop in 5-year overall survival from 74 to 96 % for stage I/II patients to 49–60 % for node positive patients is observed (Lewin et al. 2010; Werner et al. 2012). A variety of features are associated with increased risk for lymph node metastasis. The strong association of tumor grade, depth of myometrial invasion and pelvic lymph node involvement was first demonstrated in the results of GOG study 33 (Tables 8, 9) (Creasman et al. 1987). In this clinical-pathologic study, 621 stage I endometrial cancer patients, accrued from 1977 to 1983, prospectively underwent hysterectomy, selective pelvic and para-aortic lymph node dissection and peritoneal cytology. Increasing FIGO grade and increasing depth of invasion correlated with progressively higher probability of pelvic lymph node involvement, ranging from less than 5 % in patients without myometrial invasion, to 34 % in those with both outer third myometrial invasion and FIGO grade 3 histology.

Table 8 GOG 33: Rate of pelvic lymph node metastasis based on extent of myometrial invasion and FIGO grade

Table 9 GOG 33: rate of para-aortic lymph node metastasis based on extent of myometrial invasion and FIGO grade

While node positive patients as a whole have poorer survival compared to stage I and II patients, it should be noted that the predictive outcome of node positive disease should be considered in the context of the extent of other extrauterine disease. Mariani et al. examined the outcomes of 51 patients with surgically staged IIIC disease. In this cohort, it was noted that the 5-year recurrence free survival (RFS) for node positive only disease was 68 %, but dropped to 25 % in patients with node positive disease in combination with other extrauterine disease such as adnexal, vaginal, serosal involvement or positive peritoneal cytology (Mariani et al. 2002a). While this study is limited in its correlation to today’s practice as few patient received chemotherapy, this poorer outcome in “higher burden” disease suggests these patient may require a more aggressive treatment approach. The overall nodal disease burden, as described by absolute number of positive lymph nodes and ratio of positive nodes to total nodes on lympadenectomy has also been shown to be prognostic in some studies (Chan et al. 2007a). Five-year disease-specific survival for those with 1, 2–5, and >5 positive nodes were 68.1, 55.1, and 46.1 %, respectively (p < 0.001). Percentage of positive lymph nodes was also evaluated, with 5-year disease-specific survival of 77.3 to 60.7 to 40.9 % in those with ≤0, >10 to ≥50 %, and >50 % nodes involved, respectively. Both factors were independently prognostic on multivariate analysis.

3.2.3 Adnexal and Serosal Involvement

FIGO stage IIIA is defined by serosal and/or adnexal disease spread. Adnexal involvement is associated with poorer outcomes, but is highly correlated with other adverse features such as high tumor grade, other metastatic sites, and unfavorable histology. When considering adnexal involvement in the absence of other factors, outcomes are more favorable than for all stage IIIA patients taken as a whole, with 5-year disease-free survival ranging from 71 to 86 % (Connell et al. 1999; Greven et al. 1989). Serosal involvement is associated with high risk of distant failure, owing in part to its association with other risk factors such as other sites of metastatic disease and higher stage presentation (Greven et al. 1989; Ashman et al. 2001). Similar to adnexal involvement, however, isolated serosal involvement portends an improved prognosis over all patients with serosal involvement, with 5-year disease-free survival of 41.5 % versus 20 % (Ashman et al. 2001).

3.3 Patient Factors

3.3.1 Age

Age has long been considered a risk factor for development of endometrial cancer, as well as prognostic of outcomes. In general, endometrial cancer is a disease of postmenopausal women. Younger women who develop endometrial cancer tend to have improved survival, often with risk factors such as estrogen or other hormone related-disorders, including but not limited to, infertility, polycystic ovarian syndrome, ovarian dysfunction, anovulatory cycles, and obesity (Ota et al. 2005). Young patients tend to have low grade endometrioid histology, correlating to more favorable outcomes.

While many studies have shown advanced age to be an independent predictor of worse outcomes (Kosary 1994; Abeler and Kjorstad 1991; Irwin et al. 1998), many small studies have found this to not be a prognostic factor. Some of been concerned that patient comorbidities, potential de-escalation of therapy in the elderly, or narrow cohorts, or propensity for more advanced stage at diagnosis, or more aggressive histology at diagnosis, among a multitude of other confounding factors, may explain the apparent discrepancy. Regardless, age is still part of the risk stratification of patients for selection of adjuvant therapy as is discussed below.

3.3.2 Serum CA-125

CA-125 is a serum tumor marker that can readily be tested, commonly used to monitor ovarian cancer. The role of CA-125 in endometrial cancer has been proposed to be prognostic, with elevated preoperative CA-125 levels associated with increased risk of lymph node metastasis (Chung et al. 2006). Many suggest measurement of preoperative serum CA-125 given several studies suggestive of prognostic utility (Powell et al. 2005); although no change in therapy is offered based on this value. Some have also proposed an age stratified CA-125 cutoff to improve the predictive value of CA-125 levels, with higher cutoffs proposed in younger patients (Chao et al. 2013). The NCCN guidelines designate CA-125 as an optional test in both workup and surveillance, while the American Society of Gynecologists Oncologists does not endorse the routine use of CA-125 during surveillance in the absence of clinical findings concerning for metastatic disease (Salani et al. 2011). Future studies regarding the use of CA-125 are warranted and will likely focus on its potential as a tool for prediction of extrauterine disease in early stage patients or its use during surveillance for early detection of disease recurrence and whether this translates to improved patient outcomes.

3.4 Histologic Factors

While tumor stage is the most important prognostic indicator, many of the other confirmed prognostic features relate to information from the histology of the tumor itself. Tumor cell type, grade of differentiation, and LVSI are significantly important, and assist with stratification of patients within surgical staging groups into risk categories. Thus, the results of each can have significant influence on the adjuvant therapy given, as patients with early stage, low risk histology may not require adjuvant therapy, the same stage patient with high risk histology or tumor grade may have poorer outcomes if adjuvant therapy is not offered.

3.4.1 Histology

Given that surgical staging predominates for endometrial cancer, characteristics found on pathologic evaluation are highly prognostic. Cell type and tumor grade are highly predictive of patient outcomes, and carry significant weight in determining if adjuvant therapy after hysterectomy should be offered. Additional information regarding myometrial invasion, cervical stromal invasion, lymphovascular invasion, and others have also been shown to be prognostic and are used to help stratify risk of recurrence in patients with early stage disease. The following section on histology relates to histologic factors studied largely in endometrioid adenocarcinomas. In general, non-endometrioid histologies such as papillary serous and clear cell adenocarcionoma are highly correlated with many of these adverse pathologic factors, thus are deemed high risk in even early stage disease, and are offered more aggressive adjuvant therapy.

3.4.1.1 Histologic Type

The vast majority of endometrial cancers arise within the endometrial layer of the uterus, with subsequent growth and spread, usually into the myometrium, as it progresses. Adenocarcinoma accounts for the majority of endometrial cancer cases diagnosed. The most common histologic subtype is endometrioid histology, accounting for nearly 75–80 % of endometrial cancer cases. This is a gland forming variant of adenocarcinoma, often with appearance similar to that of the endometrium. Overall prognosis for low grade endometrioid adenocarcinoma is favorable. By some reports, approximately 25 % of adenocarcinomas can have squamous differentiation, where the grade of the glandular component is prognostic (Abeler and Kjorstad 1992). Villoglandular and mucinous adenocarcinomas are infrequently identified, with no significant effect on outcomes with villoglandular (Zaino et al. 1998a), and improved outcomes with mucinous features (Ross et al. 1983). Two less common, yet clinically significant subsets of adenocarcinoma, include papillary serous and clear cell adenocarcinoma, accounting for a majority of the remaining non-endometrioid cases. Papillary serous carcinomas histologically have a complex papillary architecture, resembling serous carcinoma of the ovary. Nuclear atypia is common, and psammoma bodies can be present. Clear cell carcinomas have 3 types of growth patterns, tubulocystic, papillary, or solid patterns, and are less likely to contain psammoma bodies. Any tumor that contains 10 % or more of either papillary serous or clear cell adenocarcinoma features are classified as mixed histology, although prognosis tends to correlate with the most advanced histology in the tumor.

Endometrial cancer is subdivided into type 1 or type 2 tumors; type 1 defined as low grade (FIGO grade 1 and 2) endometrioid tumors (nearly 80 % of adenocarcinoma), and type 2 encompassing FIGO grade 3 endometroid tumors, papillary serous, and clear cell adenocarcinomas. A different etiology of tumorigenesis has been proposed in these two subgroups. Type 1 tumors are generally associated with the classical risk factors for endometrial cancer including nulliparity, obesity, unopposed estrogen, early menarche/late menopause, tamoxifen therapy, among others. It has been proposed that elevated estrogenic state experienced in these situations can stimulate the endometrial layer, leading to hyperplasia, a likely precursor to endometrial cancer in some settings. Type 2 tumors, on the other hand, are not associated with hyperestrogenism or endometrial hyperplasia. Stage by stage, more aggressive histology is associated with poorer clinical outcomes (Boruta et al. 2004). As such, type 2 tumors are often included as a risk factor warranting intensification of adjuvant therapy as discussed below.

Uterine sarcomas (endometrial stromal sarcomas, leiomyosarcomas, and other mesenchymal tumors), and mixed epithelial and mesenchymal tumors (adenosarcomas and malignant mixed mullerian tumors), are much less common types of uterine cancer. As a group, they all confer very poor prognosis at diagnosis. They tend to be associated with higher stage at diagnosis, and dismal disease free and overall survival (Prat 2009; Callister et al. 2004). More aggressive therapy is generally favored in this group of patients given their significantly higher risk for failure and death, however given the relative rarity, poor response to proposed interventions, and paucity of prospective data, there is no clearly defined guideline in management (Rauh-Hain and Del Carmen 2013; Kanthan and Senger 2011).

3.4.1.2 Tumor Grade

Across a multitude of studies, tumor grade has been shown to be strongly associated with prognosis, degree of myometrial invasion, and risk for lymph node metastasis. FIGO grading of endometrioid carcinomas incorporate the degree of gland formation and nuclear grade. The percent solid (nonglandular) growth is scored as increased solid growth is associated with more aggressive behavior. Grade 1 is defined as no more than 5 % solid growth, grade 2 with 6 to 50 percent solid growth, and grade 3 with more than 50 percent solid growth. If glandular grade is different from nuclear grade, nuclear grade predominates. Non-endometrioid tumors are graded by nuclear grade alone. Zaino et al. reported 5-year survival rates of 94 % for grade 1, 84 % for grade 2, and 72 % for grade 3 tumors (Zaino et al. 1998a). Given the significant prognostic feature of tumor grade, it is incorporated into risk stratification of patients within a given stage to help direct adjuvant therapy.

3.4.1.3 Myometrial Invasion

Degree of myometrial invasion has been shown to be an independent predictor for outcome in a multitude of studies (Creasman et al. 1987; Morrow et al. 1991). This has been validated since originally described and continues to be incorporated as part of the current FIGO staging. While risk factor groups have been described based on thirds of invasion, the most recent revision of FIGO staging has established 50 % as the cutoff between stage IA and stage IB endometrial cancer.

3.4.1.4 Cervical Stromal Invasion

Cervical stromal invasion is included in FIGO staging, given its prognostic significance in outcomes with reduced 5-year disease-free survival of 74–80 % for stage II disease compared to 90–96 % for stage IA. Previously, any cervical invasion was classified as stage II disease in the 1988 FIGO schema, with stage IIA defined as isolated endocervical epithelial involvement and stage IIB for deeper stromal invasion. However, several reports failed to demonstrate a difference in survival between the two groups (Orezzoli et al. 2009; Eltabbakh and Moore 1999). Thus, this subclassification was eliminated with the recent 2009 revision of FIGO staging and currently cervical stromal invasion only constitutes stage II disease. This has been shown to be independently prognostic for patient outcomes, with a 44 % increase in risk of progression or death and a 33 % increase in risk of death (Tewari et al. 2012).

3.4.1.5 Lymphovascular Space Invasion

Lymphovascular space invasion (LVSI) has been shown to be a predictor of risk of relapse and poorer survival, independent from tumor grade or depth of myometrial involvement (Morrow et al. 1991; Mariani et al. 2002b, c). LVSI has been shown to increase the rate of pelvic lymph node metastasis (Creasman et al. 1987). LVSI continues to be used as one of several histologic criteria for risk stratification for adjuvant therapy selection and clinical trial inclusion.

3.4.1.6 Peritoneal Cytology

In previous 1988 FIGO staging, the presence of malignant cells in peritoneal fluid was designated stage IIIA disease. However, multiple studies failed to show this as an independent prognostic factor (Hirai et al. 1989; Tebeu et al. 2004; Takeshima et al. 2001). The revised 2009 FIGO staging has eliminated positive peritoneal cytology as a factor in staging. However, recently Milgrom et al. showed that in stage III patients, positive peritoneal cytology was predictive of outcome and associated with distant relapse (Milgrom et al. 2013). This is consistent with the observation that positive peritoneal cytology, while not independently prognostic, may enhance the negative impact of other adverse factors (Takeshima et al. 2001). Peritoneal cytology is still obtained at most institutions during hysterectomy, as it may have some effect on adjuvant therapy selection, and is used as inclusion criteria of some ongoing phase III trials.

3.4.2 Implications of Postoperative Histology on Adjuvant Treatment

As previously discussed, multiple histologic and clinical factors have been found to be independently prognostic of clinical outcome. While some of these are directly used for staging, others are used for risk stratification to help predict a benefit from adjuvant therapy and aid in the selection of adjuvant therapy.

3.4.2.1 Risk Group Stratification in Early Stage Endometrial Cancer

Adjuvant therapy in endometrial cancer is dictated in large part by stage and risk factors within each stage. This is specifically true for early stage endometrial cancer where the extent and method of adjuvant radiotherapy has evolved.

Observation is reasonable for patients with stage IA, grade 1, favorable histology disease, otherwise deemed low risk. In patients with stage I disease, and any risk factor, including Grade 2–3 disease, LVSI, lower uterine segment involvement, deep myometrial invasion, or advanced age > 50–70, adjuvant therapy has traditionally been considered. Previously, the GOG 33 data demonstrated advanced grade or deep myometrial invasion were risk factors for lymph node positive disease, which was associated with worse disease free survival. These risk factors had been employed to determine the need for postoperative pelvic radiation, but how to much weight to assign these risk factors has evolved.

The traditional indications for pelvic radiation in early stage disease have been challenged by the results of the Post-Operative Radiation Therapy in Endometrial Cancer (PORTEC) (Creutzberg et al. 2003, 2004) and GOG 99 (Keys et al. 2004) studies, resulting in identification of a new set of risk factors. This paradigm change has also been fueled by advances in surgical approach over the past 2 decades, with a more comprehensive degree of lymph node dissection, even in co-morbid patients. Based on both trials’ results, a new high-intermediate risk group was defined by each cooperative group and a more multi-faceted algorithm was developed that incorporated grade, depth of myometrial invasion, LVSI, and age. PORTEC’s and GOG 99’s results are highly consistent showing an incidence of failure in the 30 % range for the GOG-defined high-intermediate-risk group and for the grade 3 group in PORTEC. The high-intermediate-risk group was defined by GOG as (1) grade 2–3 with deep third myometrial invasion and LVSI; or (2) age > 50 and two of the risk factors in (1); or (3) age > 70 and one of the risk factors in (1). The definition based on the PORTEC data is similar: <50 % myometrial invasion and grade 3 (any age); or >50 % invasion and grade 1–2 and age > 60 years. However, the combination of <50 % myometrial invasion, grade 3 and LVSI is considered a high-risk feature by PORTEC-2 due to the significantly lower 5-year overall survival of 58 % observed in the PORTEC 1 study (Creutzberg et al. et al. 2004). These overall results are supported by a metaanalysis by Kong et al. (2007) of all four randomized trials (Creutzberg et al. 2000, 2004; Keys et al. 2004; Aalders et al. 1980) that shows adjuvant radiotherapy improved disease specific and overall survival for patients with grade 3 tumors and stage IB (>50 % invasion) disease. The failure pattern in the high-intermediate-risk group has been found to consist largely of vaginal recurrences, therefore, while the high risk patients are often recommended pelvic radiotherapy, high-intermediate risk group patients are often offered vaginal cuff brachytherapy and/or pelvic radiotherapy as vaginal recurrences are the most likely site of failure. This group has been studied by PORTEC-2, and vaginal cuff brachytherapy was found to be equivalent in preventing pelvic recurrence to whole pelvic radiation (Nout et al. 2010).

Adjuvant therapy for high risk disease is an area of active research as there is data to suggest intensifying therapy with chemotherapy is warranted, and currently practiced at many institutions. PORTEC-3 is currently enrolling patients with the high risk criteria and randomizing patients postoperatively to pelvic radiation or pelvic radiation with concurrent and post-radiation chemotherapy. Eligible patients include those with <50 % myometrial invasion plus grade 3 and LVSI; >50 % myometrial invasion with grade 3, or advanced endometrial cancer, including stage II–III disease, papillary serous or clear cell histologies. The results are eagerly anticipated.

3.4.2.2 Locally Advanced Endometrial Cancer

Stage III and IVA endometrial cancer is often described as locally advanced endometrial cancer. This group represents a heterogenous group of patients, with varying degrees of tumor burden and tumor histology, with the best adjuvant therapy not clearly defined. GOG 122 established a role for chemotherapy over whole abdominal radiation owing to improved disease free and overall survival of 38–50 %, and 42–55 %, respectively (Randall et al. 2006). More recently, Hogberg et al. compiled the data from two randomized European trials, NSGO-EC-9501/EORTC-55991 and MaNGO ILIADE-III, which randomized patients to adjuvant radiotherapy alone or sequential chemotherapy and radiation therapy. This indicated a significant improvement in 5-year progression-free survival from 69 to 78 %, with a trend for improved overall survival (Hogberg et al. 2010). The extent of radiotherapy, timing with chemotherapy, and patient selection is still an area of active study.

3.4.3 Molecular Markers

Molecular markers are an area of active interest. In most cases, markers are correlated with established prognostic indicators, such as tumor histology and grade. Some of the most studied factors are briefly reviewed. To date, the clinical utility of these markers is limited.

3.4.3.1 DNA Ploidy

DNA content, or more specifically, aneuploidy, has been studied by many groups. The frequency of aneuploidy has been shown to increase with increased tumor grade (Lundgren et al. 2002). Papillary serous carcinoma has been shown to exhibit aneuploidy, as well (Prat et al. 1994). Further, DNA aneuploidy has been shown to be an independent predictor for disease free survival (Zaino et al. 1998b; Nordstrom et al. 1996).

3.4.3.2 Microsatellite Instability

Microsatellite instability (MSI) is strongly associated with endometrial cancer in patient with HNPCC, occurring in nearly 75 % of such patients, and occurs in approximately 25–45 % of sporadic endometrial carcinomas. Microsatellites are short repeats of DNA that are integrated throughout the genome, and MSI is associated with deficits in DNA mismatch repair. In some studies, MSI is associated with improved clinical outcome (Maxwell et al. 2001). However, there is discrepancy in the published literature, with several reports showing no correlation with clinical outcome (Zighelboim et al. 2007; Baldinu et al. 2002), while others have shown MSI to be an independent prognostic indicator for poorer survival (Mackay et al. 2010; Nout et al. 2012; Steinbakk et al. 2011). This disagreement may be related to sample size, cohort selection, different adjuvant therapies, confounding variables, or may indicate identification of the specific downstream genetic alterations is actually more relevant (Steinbakk et al. 2011).

3.4.3.3 Ki-67 Proliferation Index

Cellular proliferation is an area of interest in most cancer cell types. This has also been evaluated by many groups for endometrial cancer. Nuclear Ki-67 antigen is a marker of proliferating cells, and has been shown to be associated with histological grade and depth of myometrial invasion, as well as other risk factors (Kudela et al. 2012). High levels of Ki-67 expression have also been associated with increased risk of recurrence and poorer survival in some studies (Salvesen et al. 1998).

3.4.3.4 Oncogenes

HER2 and EGFR are both members of the ErbB/HER signaling family, a group of tyrosine kinase receptors critical in cellular proliferation and differentiation, and are implicated in tumorigenesis in many tumor models. HER2 expression was associated with higher tumor grade and depth of myometrial invasion but not independently prognostic for survival, whereas EGFR overexpression in endometrioid adenocarcinoma decreased survival from 89 to 69 % (p < 0.04), and in serous papillary and clear cell from 86 to 27 % (p < 0.03) (Khalifa et al. 1994; Konecny et al. 2009). There is continued interest in this pathway as inhibitors of EGFR and HER2 are actively used in other cancer treatment and exploitation of this pathway with these pharmaceuticals theoretically may improve patient outcomes.

P53 has been reported to be more highly expressed in type 2 tumors (Kudela et al. 2012). Not surprisingly, this has also been correlated with poorer patient outcomes (Mariani et al. 2000; Saffari et al. 2005; Silverman et al. 2000). Currently, clinical utility of this marker is uncertain as no targeted therapies are readily available.

The evaluation of PTEN as a prognostic factor is also controversial. PTEN is a tumor suppressor gene that down regulates the PI3-Kinase pathway, thus slowing down cellular proliferation. PTEN is mutated in approximately 20–80 % of endometrial cancers, but with less frequency in serous carcinoma. Results regarding the effect of PTEN on patient outcomes is mixed (Latta and Chapman 2002).

3.4.3.5 Cell Adhesion Molecules

Cell adhesion molecules have been widely studied in tumor biology, and are responsible in part for coordinating cell–cell interaction, cellular proliferation, and metastasis. E-cadherin is a cell membrane protein that complexes with cytoplasmic B-catenin regulating cellular adhesion and growth. The loss of E-cadherin expression results in release of B-catenin, which is then able to induce a subset of genes responsible for endothelial to mesenchymal transition which is one mechanism by which tumorigenesis and metastasis is thought to occur. Loss of E-cadherin expression is commonly seen in non-endometrioid endometrial carcinoma, but occasionally in endometrioid histology (Holcomb et al. 2002; Mell et al. 2004). Although in the same pathway, B-catenin has not been found to be independently prognostic of clinical outcomes (Nout et al. 2012; Singh et al. 2011).

3.4.3.6 Steroid Receptors

Expression of estrogen receptor (ER) and progesterone receptor (PR) has been extensively examined, given hormonally directed therapy is of particular interest in patients who may not be surgical candidates or have otherwise limited treatment options. Some studies indicate ER and PR expression are associated with less aggressive tumor behavior/grade (Ferrandina et al. 2005; Geisinger et al. 1986; Kadar et al. 1993; Jeon et al. 2006). While progestins are often used in relapsed or advanced disease, a recent metaanalysis indicates there is no data at present to support its use in primary disease (Martin-Hirsch et al. 2011); prospective evaluation of receptor expression and treatment response is warranted.

3.5 Imaging Prognostic Factors

FIGO staging for endometrial cancer by definition requires surgical staging. In the United States, a majority of centers include routine pelvic lymphadenectomy and para-aortic lymph node sampling at the time of hysterectomy. Morbidity is associated with such extended surgery, although has improved with advances in surgical technology. Further, the ASTEC trial, albeit with relatively limited follow-up, to date has not shown a survival benefit to lymphadenectomy in early stage disease (ASTEC study group et al. 2009). Thus, there is great interest in developing new ways to predict risk of lymph node involvement, and to identify those patients with acceptably low risk of involvement in order to identify patients where omission of lymphadenectomy is reasonable. While clinical exam prior to 1988 was shown to understage endometrial cancer in 13–22 % of patients, newer imaging technology is now available, and may be promising in identification of factors such as myometrial invasion, extrauterine involvement, as well as risk of pelvic lymph node disease. These are briefly reviewed here.

3.5.1 Morphologic Imaging

Computed tomography (CT) has been used for preoperative assessment in endometrial cancer, but its role is with limitations. The ability of CT to delineate endometrial cancer in the uterus is relatively insensitive, especially for small endometrial cancers (i.e. stage IA), with overall sensitivity of 53 % (Grossman et al. 2008). Accuracy of CT for myometrial invasion has been reported to be 61 % with sensitivity of 40 % in one study comparing ultrasound, CT, and MRI for depth of myometrial invasion assessment (Kim et al. 1995). Multidetector CT has improved accuracy for depth of myometrial invasion and cervical involvement at 95 and 81 %, respectively (Tsili et al. 2008). The applicability of this modality is limited given this single experience in 16 patients, thus warrants further evaluation. Sensitivity and specificity of CT for lymph node involvement has been reported at 52 and 92 %, respectively (Connor et al. 2000). Chest CT can be considered in high risk patients, such as advanced stage or high grade tumors who are at increased risk for pulmonary metastasis.

The accuracy of ultrasound for myometrial invasion has been described by many groups. The accuracy of transvaginal ultrasound (TVUS) for predicting stage IA versus stage IB endometrial cancer reportedly ranges from 69 to 93 % (Kim et al. 1995; DelMaschio et al. 1993; Prompeler et al. 1994). High-frequency TVUS has been shown to have accuracy of 73 % for assessment of myometrial invasion (Arko and Takac 2000). The reported experience of ultrasonography to predict cervical involvement has also been limited, with only 7 of 10 patients with pathologic cervical involvement reported pretherapy to have involvement based on ultrasound (Akbayir et al. 2011; Szantho et al. 2001). The use of 3D ultrasonography with volume contrast imaging has also been described. Jantarasaengaram et al. reported accuracy of 92 % for predicting myometrial invasion and 90 % for cervical involvement (Jantarasaengaram et al. 2013). Sonohysterography, which involves intracavitary infusion of saline followed by evaluation with TVUS, has been employed in some settings, with accuracies of 84–89 % for assessing deep myometrial invasion (Chang et al. 2010; Valenzano et al. 2001; Dessole et al. 2006). The use of this modality is controversial, however, due to concern of tumor spillage into the peritoneal cavity with saline infusion, which has been documented by some investigators (Dessole et al. 2006; Alcazar et al. 2000).

The use of ultrasound has been compared to MRI in multiple investigations, and consistently has been found to be superior to ultrasound for evaluation of cervical involvement and depth of myometrial invasion (Kim et al. 1995; DelMaschio et al. 1993; Arko and Takac 2000; Antonsen et al. 2013a; Yamashita et al. 1993a). Further, contrast enhanced MRI, compared to unenhanced MRI, results in significantly improved accuracy, ranging from 85 to 92 % accuracy for depth of myometrial invasion versus 55–78 % for non-contrasted imaging (Kinkel et al. 1999; Ito et al. 1994; Saez et al. 2000; Sironi et al. 1992; Yamashita et al. 1993b; Sala et al. 2009). Accuracy rates for determination of cervical involvement range from 86 to 95 % (Manfredi et al. 2004; Takahashi et al. 1995; Nagar et al. 2006). The use of MRI for pelvic and para-aortic lymph node involvement is comparable to CT, with sensitivity and specificity reported at 44–66 % and 73–98 %, respectively. Thus, given MRI’s superior assessment of depth of myometrial invasion and cervical involvement, it is generally preferred over CT and ultrasound for preoperative workup.

3.5.2 Functional Imaging

The use of PET/CT in endometrial cancer is an area of active investigation. A recent meta-analysis of 18F-FDG PET or PET/CT for identification of metastatic lymph nodes in endometrial cancer reported the pooled estimates for 243 patients, indicating sensitivity and specificity of 63 % (95 % CI, 48.7–75.7 %) and 94.7 % (95 % CI, 90.4–97.4 %), respectively (Chang et al. 2012). The relatively low sensitivity is uncertain, but may be related to low glucose metabolism in low grade lesions, as well as limited ability to detect subcentimeter metastases. Further, PET imaging is limited in ability to detect intraperitoneal tumor implants and parenchymal implants. Due to these limitations, CT and MRI are preferable for detection of extrauterine disease, although FDG-PET may be appropriate in patients with high grade tumor that is likely to be FDG avid (Lee et al. 2011).

The role for PET/CT for assessment of myometrial invasion and cervical invasion is uncertain. Antonsen et al. recently reported the results of 318 patients with endometrial cancer who preoperatively underwent 2D ultrasonography, MRI, and PET/CT imaging. Sensitivity, specificity, and accuracy for PET/CT for myometrial invasion were 93, 49, and 61 %, and 43, 94, and 83 %, respectively for cervical invasion, which were similar to MRI (Antonsen et al. 2013a).

SUVmax has been evaluated by some groups, with limited data suggesting SUVmax may be able to predict higher stage disease, higher grade tumors, risk of deeper myometrial invasion, and lymph node metastatic risk (Antonsen et al. 2013b; Nakamura et al. 2010). Other studies have indicated SUVmax can also predict for poor disease free survival (Kitajima et al. 2012) and overall survival (Nakamura et al. 2011, 2013).

Finally, 18F-FDG PET or PET/CT has also been used for detection of recurrent disease (Park et al. 2008; Belhocine et al. 2002; Chung et al. 2008; Kitajima et al. 2008). Saga et al. assessed the use of 18F-FDG PET in 21 patients for detection of recurrence and evaluation of treatment response. Compared to conventional imaging and serum tumor markers, FDG-PET combined with CT or MRI was more accurate and had comparable or better sensitivity and specificity (Saga et al. 2003). Currently, the ACR guidelines indicate that FDG-PET is usually appropriate over MRI pelvis or CT pelvis if recurrence is suspected clinically (Lee et al. 2011).

4 Cancer of the Vulva

Vulvar cancer is a rare disease, accounting for only 5 % of malignancies of the female genital tract (Siegel et al. 2012). It is estimated that in 2013 there will be approximately 4,700 new cases and 900 deaths due to this disease in the United States (Siegel et al. 2012). The mean age at diagnosis for vulvar cancer is 65 years, and clinical risk factors for this disease include immunodeficiency, prior history of cervical cancer, cigarette smoking, vulvar dystrophy, vulvar or cervical intraepithelial neoplasia, and HPV infection (Ansink 1996; Madsen et al. 2008).

Vulvar cancer is a disease of the skin, arising from squamous epithelium, and tumor spread occurs primarily through the lymphatic system. The first station of nodal spread is the inguino-femoral lymph nodes, usually superficial first then deep, which then spreads in a predictable fashion to the pelvic lymph nodes in more advanced cases. Pelvic lymph node involvement without inguinal node involvement is rare (Krupp and Bohm 1978). Locally, vulvar cancer can invade adjacent structures including the vagina, bladder, anus and rectum. Given the propensity of this type of cancer to spread to adjacent structures and metastasize to lymph nodes, standard of care had previously been en bloc resection of the primary tumor with inguinofemoral lymph node dissection, resulting in significant risk of morbidity and psychosexual impact. However, the approach to treatment has evolved over the last several decades, with therapy ranging from wide local excision for small, superficial lesions, to definitive or neoadjuvant chemo-radiation which may reduce the extent of surgical resection required, versus pelvic exenteration in advanced disease.

4.1 Staging

Prognostic factors for vulvar cancer include size and local extension of the primary tumor, as well as the degree of lymphatic involvement, as reflected in the most recent (2009) version of the FIGO staging system (Hacker 2009), Table 10. As with other gynecologic malignancies, the FIGO staging system is a clinicopathologic staging system and formal recommendations for the staging evaluation for vulvar cancer have not been established. The extent and size of the primary tumor is established by clinical examination, often by EUA, including colposcopy, excisional biopsy or FNA of clinically positive inguinal nodes, and/or cystoscopy and proctoscopy based on presentation in advanced disease. Clinical palpation alone does not have a high degree of specificity or sensitivity for inguino-femoral adenopathy (Homesley et al. 1993; Franklin 1972; Selman et al. 2005). Thus, imaging modalities such as MRI and PET/CT, as well as CT of the chest, abdomen, and pelvis are also typically employed, particularly to evaluate for lymph node involvement. Several studies have shown that MRI may be useful in evaluating the inguinofemoral lymph nodes (Singh et al. 2006; Sohaib et al. 2002). However, it is important to note that radiologic findings cannot be used as a substitute for pathologic assessment of the nodes.

Table 10 AJCC TNM and FIGO staging of vulvar cancer

The gold standard for pathologic assessment of the inguinofemoral lymph nodes is lymphadenectomy. However, this carries a significant risk for morbidity. Recently, studies have evaluated the utility of sentinel lymph node biopsy, rather than lymphadenctomy in select patients as this technique carries less morbidity (Hefler et al. 2008). Comparison of sentinel lymph node biopsy to lymphadenectomy in a phase II GOG study showed that a sentinel lymph node can be found in 92 % of patients, and is 92 % sensitive, with a false negative rate of 2 % in patients with tumor less than 4 cm in size (Levenback et al. 2012). Information from sentinel lymphadenectomy has not yet been incorporated into the staging system. However, data from the recent GROINSS-V study indicates that the disease burden identified in the sentinel node is a sensitive indicator of prognosis (Oonk et al. 2010).

4.2 Clinical Factors

Lymph node involvement and size and extent of primary tumor are the strongest prognostic indicators in vulvar cancer, thus the 2009 FIGO staging system incorporates both factors. In the past, wide local excision or en bloc resection with bilateral inguinofemoral lymph node dissection was the standard surgical approach. However, bilateral lymphadenectomy carries significant risk for morbidity, both in the short and long term due to wound complications, infection, and lymphedema. Therefore, efforts are made to identify a cohort of patients that may not require lymph node dissection, albeit with an abundance of caution. In early vulvar cancer, appropriate management of the lymph nodes is the single most important factor in decreasing mortality as recurrence in the undissected inguino-femoral lymph nodes results in higher mortality (Cormio et al. 2010). Tumor size and depth of invasion can help predict for risk of lymph node involvement, therefore, combined with clinical exam and imaging, the decision for surgical evaluation of lymph nodes is determined.

4.2.1 Stage

In the 1988 FIGO staging system, prognosis was well distributed among the stage categories with 89 % 5-year survival for stage I, 85 % for stage II, 74 % for stage III and 31 % for stage IV (Homesley et al. 1991). However stage III consisted of a heterogeneous group of patients, with survival ranging from 30 to 100 %. The current revised 2009 FIGO staging system improves upon the prior staging schema by including more detailed information regarding the extent of lymph node involvement by subdividing the category into Stage IIIA and IIIB based on size and number of lymph nodes, or extracapsular spread (Stage IIIC), all of which are reported to closely correlate with prognosis (Homesley et al. 1991; Hacker et al. 1983; Lataifeh et al. 2004; Origoni et al. 1992; Raspagliesi et al. 2006; Fons et al. 2009a; Woelber et al. 2009). Several recent studies have validated the prognostic utility of these expanded stage categories (Tan et al. 2012; Tabbaa et al. 2012).

4.2.2 Tumor Volume

While lymph node involvement is the most important outcome predictor in multivariate analysis, tumor size has been shown to be an independent prognostic factor for local recurrence (Rutledge et al. 1970). However, in several larger contemporary series, tumor size does not independently predict disease free and overall survival on multivariate analysis (Raspagliesi et al. 2006; Tantipalakorn et al. 2009), although associations between T-stage and local recurrence-free and disease-free survival are seen in univariate analysis in a large series of 215 patients with reported local recurrence-free survival and disease-free survival rates of 85 and 88 % in T1 lesions, 74 and 61 % in T2 and 69 % and 37 % for T3/4 tumors (Rouzier et al. 2002). Finally, tumor size correlates closely with the probability of lymph node involvement, thereby likely conferring worse prognosis through its association with this unfavorable risk factor. The incidence of lymph node involvement is 5–8 % for tumors <1 cm and increases to 24 % for 1–2 cm, 31 % for 2–3 cm and 36 % for 3–5 cm tumors (Gonzalez Bosquet et al. 2003; Boyce et al. 1985).

4.3 Patient Factors

4.3.1 Age

While the mean age at diagnosis of vulvar cancer is 65, advanced age is a prognostic factor for increased risk of groin node metastasis and worse survival (Homesley et al. 1993; Sznurkowski et al. 2013; Blecharz et al. 2008; Ramanah et al. 2012). However, this variable has not been significant in some cohorts when adjusted for stage, lymph node status, and surgical therapy (Raspagliesi et al. 2006; Woelber et al. 2009; Burger et al. 1995).

4.3.2 Hemoglobin

As in many cancers, tumor hypoxia is thought to be one factor for poor prognosis. Vulvar cancer patients with anemia have been found to have higher incidence of inguinal lymph node metastasis (Stone et al. 2005; van de Nieuwenhof et al. 2010). On univariate analysis, Hefler et al. showed hemoglobin <12 g/dl resulted in shorter survival, similar to van de Nieuwenhof et al. that showed hemoglobin <11.3 g/dl was an independent predictor of poorer survival (van de Nieuwenhof et al. 2010; Hefler et al. 2000). Interestingly, this did not correlate with expression of hypoxia markers GLUT-1 nor CA-IX. Further evaluation of this variable has not been studied, therefore clinical utility is uncertain, as anemia may simply be a marker of poorer overall health.

4.4 Histologic Factors

4.4.1 Histology

Nearly 85–90 % of vulvar malignancies are squamous cell carcinoma. Melanoma is the second most common while other histologies such as basal cell carcinoma, Bartholin’s gland adenocarcinoma, Merkel cell and sarcomas are more rare (Hunter 1975; Finan and Barre 2003; Sugiyama et al. 2007; Stang et al. 2005; Ragnarsson-Olding et al. 1993; Weinstock 1994). Squamous cell carcinoma are often classified into one of two types; classic, warty, Bowenoid type or keratinizing, differentiated, simplex type. Squamous cell carcinoma of the vulva often arise in the setting of premalignant conditions such as vulvar intraepithelial neoplasia (VIN) or other areas of chronic inflammation such as lichen sclerosis Bowen’s disease, Paget’s disease, and erythroplasia of Queyrat (Carlson et al. 1998; Kutlubay et al. 2013). A large subset of premalignant conditions, particularly usual type VIN, are associated with HPV infection and are more likely to be observed in young women or smokers and warty or basaloid type squamous cell tumors. Conversely, keratinizing type is more often reported in older women in the setting of chronic inflammation such as lichen sclerosis (Hildesheim et al. 1997; de Koning et al. 2008; Del Pino et al. 2013). This dichotomy is thought to be a result of two different tumorigenic mechanisms that can result in vulvar squamous cell carcinoma, HPV dependent or independent mechanisms. Several reports indicate patients with HPV positive tumors have better survival than those with HPV negative tumors (Lindell et al. 2010).

Data supporting prognostic implications of tumor grade or LVSI is varied. In multiple studies, higher tumor grade is associated with increased risk for lymph node metastasis and worse overall survival (Homesley et al. 1993; Sznurkowski et al. 2013; Podratz et al. 1983; Lavie et al. 1999). However, it was not a significant variable for survival according to Burger et al. or Lataifeh et al. (Lataifeh et al. 2004; Burger et al. 1995). LVSI is associated with increased risk of lymph node metastasis (Homesley et al. 1993; Husseinzadeh et al. 1990; Binder et al. 1990) and is significant for overall survival on univariate analysis (Lataifeh et al. 2004; Raspagliesi et al. 2006; Burger et al. 1995; Knopp et al. 2004; Paladini et al. 1994), but only retains significance on multivariate analysis in a select few reports (Raspagliesi et al. 2006; Knopp et al. 2004).

4.4.2 Depth of Invasion

Depth of tumor invasion, defined as the distance from the epithelial/stromal junction to the deepest point of invasion, correlates strongly with lymph node involvement. While the risk of lymph node involvement for tumors with < 1 mm invasion is essentially nil, it increases to 6 % for 1–2 mm depth of invasion, 8 % for 2–3 mm, 22 % for 3–4 mm, 25 % for 4–5 mm and 38 % for > 5 mm depth of invasion. Lymph node dissection is therefore recommended for tumors with a depth of invasion of >1 mm (Berek and Hacker 1989). Similar criteria should be applied for the decision of adjuvant radiation in un-dissected groins. Other investigators observed variable threshold level of 3 mm (Woelber et al. 2009) and 9 mm as predictors of relapse and survival (Nicoletto et al. 2010).

4.4.3 Surgical Margins

A clear association between surgical margins and local failure has been shown. Microscopic margins of <8 mm (in formalin fixed tissue) are associated with a local recurrence rate of 48 %, compared to no recurrences with wider margins (Heaps et al. 1990). This correlation has been substantiated in a more recent study showing a 23 % incidence of local recurrence in patients with margin distance of < 8 mm, compared to no recurrences in those with >8 mm margins (Chan et al. 2007b). Adjuvant radiotherapy significantly reduces local recurrence rates for both close and positive margins and improved survival (Faul et al. 1997; Viswanathan et al. 2013).

4.4.4 Histopathologic Lymph Node Status

Lymph node status is the single most significant prognostic factor in vulvar cancer. Lymph node positivity has a profoundly adverse effect on treatment outcome, with survival declining from >90 % in patients with negative lymph nodes to as low as 30 % or less in those with involved lymph nodes (Homesley et al. 1993; Hacker et al. 1983; Origoni et al. 1992; Rutledge et al. 1970; Podratz et al. 1983; Chan et al. 2007b; Iversen et al. 1980).

Number and pathologic extent of the lymph node involvement are of paramount importance for prognosis (Table 11). Patients with involvement of one lymph node and small primary tumors tend to have a survival above 90 %, whereas survival is reduced below 35 % in those with 2 or more nodes (Homesley et al. 1991). In a large single institution study of 389 patients, nodal status was the most significant independent prognostic factor, followed by LVSI. Within the node-positive group, percentage of nodal replacement and extracapsular spread independently predicted outcome (Raspagliesi et al. 2006).

Table 11 The effect of positive lymph nodes on cancer specific survival

The prognostic significance of bilaterality of LN involvement has remained controversial, some suggesting that it influences outcome (Burger et al. 1995; Fons et al. 2009b), while others find no correlation when the number of lymph nodes is also considered (Hacker et al. 1983; Raspagliesi et al. 2006). Of note, the most recent revision of the FIGO staging has eliminated laterality of lymph node involvement in favor or number and size, and the presence or absence of extracapsular extension.

Recommendations regarding adjuvant therapy have been informed in part by GOG 37, in which 114 patients who underwent radical vulvectomy and bilateral inguino-femoral lymph node dissection, and with positive lymph nodes were randomized to adjuvant bilateral inguinal and pelvic radiation versus pelvic node dissection (Homesley et al. 1986). Patients with >2 involved inguinal nodes showed a significant benefit from adjuvant radiotherapy over surgery. The study was underpowered to draw clear conclusions on involvement of one or two lymph nodes. However, single-institution studies suggest that gross involvement of a single node also has a substantial recurrence risk, particularly if extranodal extension is present, and warrants consideration of adjuvant therapy (Origoni et al. 1992; Ansink et al. 1991).

4.4.5 Molecular Markers

4.4.5.1 DNA Ploidy

Aneuploidy has been reported to correlate with other poor prognostic factors, and has been reported to predict for worse outcome (Lerma et al. 1999; Mariani et al. 1998), although other studies have shown no significant relationship (Knopp et al. 2004; Dolan et al. 1993).

4.4.5.2 HPV Dependence and Independence

HPV positive tumors, which occurs more commonly in younger patients, may be associated with a better prognosis (Monk et al. 1995; Ansink et al. 1994). Given that vulvar cancer is thought to be driven by HPV-dependent and independent pathways, several groups have looked at associated markers. Basaloid and warty tumors, often considered HPV-dependent tumors, often express p16, and are p53 negative, whereas keratinized tumors, classically HPV-independent tumors, are p16 negative and p53 positive (Santos et al. 2004; Kruse et al. 2008). Investigators have also explored if markers of HPV infection, such as p16^INK4a, a protein that is increased due to HPV E7 oncogene activity, is prognostic in vulvar cancer. Knopp et al. and Tringler et al. showed high expression was associated with improved survival on univariate, but not multivariate analysis (Knopp et al. 2004; Tringler et al. 2007). The expression of p53 has also been associated with poorer overall survival in several studies (Hoffmann et al. 1999; Scheistroen et al. 1999; Kohlberger et al. 1995). Of note, in Scheistrøen et al., this was only found in stage III vulvar cancer, and not stage I and II disease. Kagie et al. and McConnell et al. did not find p53 overexpression to be a significant prognostic indicator, but did note its presence in adjoining premalignant lesions, such as VIN, perhaps indicating it as a marker for malignant transformation from precursor lesions (Kagie et al. 1997; McConnell et al. 1997).

4.4.5.3 ErbB/HER Signaling Family

HER2 and EGFR expression have also been identified in a variety of small studies as prognostic indicators for clinical outcome. HER2 and EGFR overexpression has been identified in 47 and 67 % of vulvar cancers, respectively (Hantschmann et al. 2005; Johnson et al. 1997). Further, both HER2 and EGFR expression has been associated with increased risk for lymph node metastasis, while EGFR overexpression, in the absence of HPV infection, is associated with decreased survival (Johnson et al. 1997; Gordinier et al. 1997; Woelber et al. 2012; Growdon et al. 2008). Given that small molecule EGFR tyrosine kinase inhibitors and HER2 directed therapies are available, this may represent a cohort of patients that may benefit from targeted therapies in the future.

4.4.5.4 Angiogenic Factors

Increased VEGF expression is associated with increased microvessel density, and has been associated with poorer survival (Jach et al. 2011; Obermair et al. 1996). CA IX, often associated with hypoxia, is up-regulated in various solid tumors, including vulvar cancer. High intratumoral expression has been associated with unfavorable disease-free survival (Kock et al. 2011; Choschzick et al. 2010). Interestingly, higher serum CA IX preoperatively was also associated with unfavorable prognosis (HR 7.2 p = 0.02) (Kock et al. 2011). The clinical utility of such measurements is uncertain.

4.4.5.5 Microarray Identified Factors

Several groups have employed microarray techniques to try to identify prognostic markers, or therapeutic targets, in vulvar cancer (Kowalewska et al. 2012; Fons et al. 2007), some of which were significantly associated with worse disease free survival including cyclooxygenase 2 and Caspase 3, (Fons et al. 2007) and SFN, CA12 and JUP which are associated with increased nodal recurrence risk and earlier time to recurrence (Kowalewska et al. 2012). However, given the limited number of cases, and unknown mechanism of these markers, further studies will be needed in order to verify any prognostic or therapeutic potential.

4.5 Treatment Related Factors

Vulvar cancer is primarily surgically treated disease, while radiation therapy plays a major role in adjuvant therapy and in locally advanced unresectable disease. Over time, the surgical approach has become more tailored toward clinical stage at presentation, with wide local excision acceptable for stage IA lesions, and the use of sentinel lymph node biopsy instead of lymphadenectomy in lateralized, clinically negative, early stage patients at the time of primary treatment with vulvectomy result in reduced morbidity without compromise in local control (Levenback et al. 2012; Van der Zee et al. 2008).

The approach to locally advanced disease has also evolved. More recently, the use of neoadjuvant chemotherapy and/or radiation or definitive chemoradiotherapy have been investigated. Patients with locally advanced unresectable vulvar cancer treated on the GOG 101 and 205 studies received neoadjuvant concurrent radiation and chemotherapy prior to resection of residual disease (Moore et al. 2012; Montana et al. 2000). Response to neoadjuvant therapy was a powerful predictor of local control and survival. In GOG 205, among the patients who completed therapy, 64 % achieved a complete clinical response and 50 % of patients achieved a complete pathological response (Moore et al. 2012). Among those with pathological response, local control was 75 % (22/29), and 3 local failures were salvageable with surgical resection; thus 25/29 patients with complete pathologic response are disease free. Conversely, only 43 % (9/21) of patients with incomplete response survived. Among those who did not undergo resection of persistent tumor, none survived.

4.6 Imaging Prognostic Factors

As treatment of vulvar cancer has evolved from radical vulvectomy and bilateral inguino-femoral lymphadenectomy to a more tailored surgical and neoadjuvant chemoradiation approach, the accuracy of pre-treatment staging is increasingly important.

Imaging prognostic factors have not been clearly identified in vulvar cancer. However molecular imaging is emerging as a useful tool to identify lymph node involvement, location and extent. Given the rarity of vulvar cancer, diagnostic imaging utility is extrapolated from experience in cervical and anal cancer. In a small prospective study, PET/CT has been shown to have a sensitivity of 67 % and specificity of 95 % in indentifying lymph node involvement, and was particularly useful in detecting extranodal involvement (Cohn et al. 2002), which all constitute powerful prognostic factors. Thus patients may be triaged to more aggressive therapy based on the imaging findings, however at present, imaging cannot substitute for histologic information obtained with invasive lymph node evaluation.

MRI has also become common for evaluation of vulvar cancer at diagnosis. While early stage vulvar cancers can often be staged on clinical exam, the extent of involvement of adjacent structures may be more difficult in locally advanced disease. MRI has been shown to be 70–85 % accurate with particular utility in defining the extent of invasion of adjacent structures and outlining tumor size, thus aiding in pretreatment surgical or radiotherapeutic planning (Kataoka et al. 2010; Sohaib et al. 2002).

5 Cancer of the Vagina

Primary vaginal cancer, defined as a lesion arising from the vagina, without involvement of the vulva or cervix, is a rare entity comprising only 1–2 % of gynecologic malignancies. The incidence of invasive vaginal cancer in the US has been reported at 0.69 per 100,000 women, with approximately 1,100 invasive cases annually. The median age at diagnosis is 68 years (Wu et al. 2008). Greater than 90 % are of squamous cell etiology. Risk factors for vaginal carcinoma include history of HPV infection, cervical intraepithelial neoplasia, prior hysterectomy, first intercourse before 17 years of age, five or more sexual partners, genital warts, chronic irritant vaginitis, and immunosuppression (Daling et al. 2002; Okagaki et al. 1983; Brinton et al. 1990; Bouma et al. 1994; Sillman et al. 1997).

Vaginal cancer is often found to be multifocal and often arises in the upper vagina. Tumor spread can be by local extension, lymphatic spread or hematogenous dissemination. Current FIGO staging is by clinical exam, chest and skeletal radiography. By definition, vaginal cancer cannot involve the vulva or cervix, therefore, multiple biopsies are performed to rule out involvement as this may change the diagnosis of the primary lesion. The lymphatic drainage of the vagina is very complex, with the upper vagina draining primarily via cervical lymphatics to the interiliac and parametrial nodes. The posterior vagina drains into the presacral, anorectal and inferior gluteal nodes, while the distal vagina drains in a vulvar pattern to the inguinal and femoral nodes, and subsequently to the pelvic nodes. Thus, all regional nodal stations are at risk for spread in vaginal carcinoma within the mid vagina, or tumors spanning several areas of the vagina.

Given the rarity of this disease, phase III trials have not been conducted, with guidelines drawn from retrospective studies and extrapolated from cervical and anal cancer experience given similarities in histology and preference for organ preservation. Similarly, prognostic and predictive factors are more challenging to elucidate in vaginal cancer due to limited data and relatively non-standardized treatment.

5.1 Staging

FIGO staging is the major prognostic indicator of disease outcome (2009). A thorough bimanual and rectovaginal exam is the most important tool for evaluation of local extent of disease, and often is carried out under anesthesia at which time biopsies can also be performed. Clinical exam focuses on differentiating vaginal wall only (stage I), extension to subvaginal tissue (stage II), or extension to the pelvic wall (stage III). In advanced disease, cystoscopy, proctoscopy, and IV pyelogram to rule out hydronephrosis may be indicated to rule out direct extension of tumor which would constitute stage IVA disease. Biopsies of the cervix or any other suspicious lesions should be performed to rule out cervical, urethral, or vulvar primaries, as these must be excluded for the diagnosis of vaginal cancer by FIGO criteria. Chest and skeletal radiography are also allowed. The results of biopsy or fine-needle aspiration of the inguinal/femoral or other nodes may be included in the clinical staging, although FIGO does not specify staging stratification for lymph node positive disease.

5.2 Clinical Prognostic Factors

5.2.1 Stage

Clinical stage is the major prognostic factor for overall survival. Most patients, except those with very limited involvement, are treated with primary radiation therapy. Based on NCDB data, one of the largest retrospective reviews of survival by stage, 5-year overall survival was 73 % for stage I, 58 % for stage II, and 36 % for stage III–IV (Creasman and Menck 1998). Similarly, in a SEER analysis by Shah et al., 5 year disease specific survival was 84 % for stage I tumors, 75 % for stage II tumors, and 57 % for stage III/IV (Shah et al. 2009).

5.2.2 Tumor Volume

Tumor size is an important predictor of outcome. In one of the largest series of patients treated with primary radiation, pelvic control was 85 % in tumors <4 cm versus 75 % in those >4 cm, and disease-specific survival was 82 and 60 % respectively (Frank et al. 2005). In a series of 301 patients by Chyle et al., lesions <5 cm maximum dimension had a 10-year local recurrence rate of 20 % compared to 40 % for >5 cm, which was significant on univariate analysis (Chyle et al. 1996). Perez et al. demonstrated tumor size was only predictive of pelvic control and disease free survival in stage II patients without parametrial involvement (Perez et al. 1999). Length of vaginal involvement has been implicated as an adverse prognostic factor (Kirkbride et al. 1995), which may also be linked to tumor size.

5.2.3 Tumor Location

While location of the tumor is important, particularly in consideration of nodal regions at risk, the utility of location as a prognostic factor is unclear. Several studies have indicated better survival and decreased risk for recurrence for patients with tumors located in the proximal half compared to those of the entire vagina or distal portion (Chyle et al. 1996; Kucera and Vavra 1991; Urbanski et al. 1996; Ali et al. 1996). Lesions of the posterior vagina wall also have worse prognosis than other locations (Chyle et al. 1996; Dixit et al. 1993). Counter to these studies, however, Perez et al. did not show any prognostic value to tumor location in the posterior vaginal wall (Perez et al. 1988).

5.2.4 Lymph Node Status

Surgical series have reported rates of pathologic nodal involvement that range from 6 %-14 % for stage I disease and 26–32 % for stage II disease (Al-Kurdi and Monaghan 1981; Davis et al. 1991). Al-Kurdi and Monaghan noted that 12 % survived when pelvic or inguinal lymph nodes were involved as compared to 47 % survival in node negative patients. However, these are small studies, and given the fact FIGO staging does not include lymph node disease, it is difficult to assess its prognostic utility.

5.3 Histologic Factors

5.3.1 Histology

The majority of invasive vaginal cancers are squamous cell histology at >90 %. Approximately 5 % of primaries are adenocarcinoma, most commonly clear cell adenocarcinoma, and 3–5 % are malignant melanoma. Other less common histologies include sarcomas, lymphoma, leukemia, and neuroendocrine small cell.

Among the histopathologic factors, correlation between histologic tumor grade and outcome has been controversial although two studies have demonstrated increased rates of recurrence with higher tumor grade (Chyle et al. 1996; Vavra et al. 1991). Adenocarcinaomas appear to confer a less favorable prognosis than squamous cell carcinomas, particularly those unrelated to DES exposure (Frank et al. 2007). DES induced clear cell adenocarcinomas in younger women arising from in utero exposure have a generally better prognosis than squamous cell carcinoma, but are unlikely to be seen today, as the use of DES during pregnancy has been banned in 1975.

5.3.2 Molecular Markers

Squamous cell carcinoma of the vagina, much like other gynecologic malignancies, are associated with HPV-dependent and independent pathways. HPV-negative tumors tend to occur in older women, with classical keratinizing, verrucous features. HPV-positive tumors are associated with basaloid, non-keratinizing lesions, tend to occur in younger patients, and present with earlier stage disease (Daling et al. 2002; Larsson et al. 2013). Larsson et al. showed significantly improved 5 year overall survival in HPV-positive tumors compared to HPV-negative tumors at 51.1 and 10.7 %, respectively (p = 0.0008). Owing to the rarity of vaginal cancer, evaluation of other molecular markers has been limited.

5.4 Treatment Related Factors

Stage is an important determinant of therapy selection, as stage I patients and selected stage II patients may be amenable to surgical therapy alone, ranging from wide local excision to vaginectomy with reported survival rates of 90 % or greater (Creasman and Menck 1998). Some population based studies have also reported improved survival with surgery over radiotherapy for early stage disease of 90 % versus 38 %, but is likely confounded due to patient selection bias and the use of radiotherapy in patients deemed poor surgical candidates due to comorbid disease (Creasman and Menck 1998). Similar results were observed in a SEER analysis that identified an adjusted hazard ratio of 1.5 for increased mortality risk in stage I patients undergoing radiotherapy in place of definitive surgery (Shah et al. 2009).

Given that organ preservation is desirable if outcomes are equivalent to surgery, the utility of definitive radiotherapy has also been explored. Superficial stage I tumors can be treated with brachytherapy alone. However several investigators have observed higher local recurrence rates in stage I patients with infiltrating lesions or higher grade tumors, thus external beam radiation has been advocated in such patients (Nori et al. 1983; Leung 1993). Definitive radiotherapy for early stage disease has reported cause specific survival of 40–90 % for stage I and 35–78 % for stage II disease (Frank et al. 2005; Perez et al. 1999; Kucera and Vavra 1991; Urbanski et al. 1996; Kirkbride et al. 1995; de Crevoisier et al. 2007; Tran et al. 2007; Prempree and Amornmarn 1985; Pingley et al. 2000). In more advanced disease, the combination of external beam radiotherapy with brachytherapy has been shown to improve pelvic control and survival in stage II vaginal cancer (Pingley et al. 2000). Locally advanced disease is often approached with external beam radiotherapy with or without chemotherapy, given to the extremely morbid exenteration that would be necessary to remove surgically.

Although the influence of tumor location within the vagina on prognosis has not been substantiated (Chyle et al. 1996; Perez et al. 1999; Kirkbride et al. 1995), tumor location has profound impact of radiation therapy planning, and treatment algorithms have been based on location and depth of invasion to optimize therapy and minimize toxicity. Following external beam therapy, upper vaginal and apical tumors are treated with intracavitary radiation, if their residual thickness is 5 mm but require more invasive interstitial therapy for thicker lesions. Anterior mid-vaginal lesions are treated with interstitial therapy, and posterior mid-vaginal lesions with a highly conformal or IMRT external beam boost due to the poor tolerance of interstitial therapy in the perirectal region. Confined distal vaginal lesions can be treated with interstitial brachytherapy, whereas massive lesions require external beam boost (Frank et al. 2005).

The role of addition of chemotherapy is still evolving. No randomized trials evaluating radiotherapy with or without chemotherapy have been performed, although chemotherapy is used concurrently in some settings given the experience of improved outcomes in locally advanced cervical cancer (Morris et al. 1999; Lanciano et al. 2005; Rose et al. 1999). Feasibility has been demonstrated in several small institutional studies. Concurrent 5-fluorouracil (5-FU) with bolus cisplatin or mitomycin C in patients with early stage disease resulted in 93 % cause specific survival at 5 years (Dalrymple et al. 2004). Five year cause specific survival of 50 % and pelvic control rate of 31 % were reported in 26 locally advanced patients treated with definitive radiotherapy and concurrent 5-FU and mitomycin C or single agent cisplatin (Kirkbride et al. 1995). A small series reporting outcomes for neoadjuvant paclitaxel and cisplatin prior to radical surgery in 11 stage II patients resulted in 27 % complete clinical response and 64 % partial clinical response with chemotherapy, and 18 % rate of disease recurrence at median follow up of 75 months (Benedetti Panici et al. 2008).

5.5 Imaging Prognostic Factors

Delineation of tumor size and degree of infiltration or spread is of critical importance, especially in the setting of definitive radiotherapy. Therefore, information obtained from radiographic workup guides treatment approach and delivery.

Given the inaccuracy of clinical palpation findings, MRI is a very useful tool for delineating extent of disease. MRI can be used in determining tumor thickness and paravaginal infiltration on T2-weighted imaging, identified as hyper-intense lesions (Taylor et al. 2007). This is in line with the superior soft tissue resolution described with MRI from cervical cancer literature Bipat et al. 2003; Hricak et al. 2005. Visualization of the vaginal tumor may be improved with the instillation of vaginal gel or a dry vaginal tampon (Young et al. 2012).

Similar to its utility in cervical cancer, PET has shown high sensitivity in identifying inguinal or pelvic lymph node involvement in advanced vaginal cancer and by some reports is more accurate than CT scan (Lamoreaux et al. 2005). In practice, PET can be critical for defining target volumes for accurate external beam and brachytherapy planning. Taken together, MRI and/or PET imaging is often obtained to identify tumor size and predict lymph node involvement and MRI is often recommended in the setting of surveillance to distinguish between tumor recurrence or radiation change. These modalities appear more sensitive than CT alone, therefore 3D CT information is often used primarly for radiotherapy treatment planning.

6 Summary

Gynecologic malignancies are somewhat unique in regard to the methods of staging and treatment compared to more commonly encountered cancers such as lung, breast, or prostate. Clinical and surgical staging have long dominated how gynecologic cancers are evaluated and substratified. With the advent of new imaging modalities and molecular diagnostic abilities, more information is available prior to selection of therapy, and the prognostic utility of these factors is evolving. As more research focuses on validating prognostic utility of imaging, histopathologic characteristics, and molecular footprints, treatment approach will likely continue to evolve.