An estimated 210,000 new cases of renal cell carcinoma (RCC) are diagnosed annually worldwide, accounting for nearly 2% of all adult cancers globally [1]. Within the International Society of Urological Pathology Vancouver classification [2], the most common histologic subtypes of renal cell carcinoma are clear cell, papillary, and chromophobe, accounting for 70%–80%, 14%–17%, and 4%–8% of RCCs, respectively [3, 4]. Of these common subtypes, clear cell RCC has the worst prognosis with a 5-year survival rate of 44%–69%, accounting for 94% of metastatic RCC [5,6,7,8,9]. Postoperatively determined pathologic stage and grade have been traditionally utilized to predict patient prognosis in RCC. However, these methods are invasive and may be unreliable due to heterogeneity within patient groups. As an alternative to the traditional approach, molecular biomarker analyses have recently been studied to address these shortcomings.

Identifying imaging biomarkers to predict specific gene and protein expression has the potential to be of great clinical value. Molecular analyses are becoming an increasingly important part of patient care, and imaging studies will soon need to provide information beyond the size and location of individual lesions. Within this new paradigm of personalized medicine, imaging features may help predict or correlate with a variety of histologic, cytologic, and molecular characteristics of individual lesions and may potentially help to guide therapy.

Multiple recent studies have shown that the magnitude of enhancement on multiphasic multidetector computed tomography (MDCT) can help distinguish clear cell RCC from other common RCC subtypes (papillary and chromophobe) and common benign mimics (oncocytomas and lipid-poor angiomyolipomas) [10,11,12,13,14,15,16,17]. Clear cell RCCs typically exhibit peak enhancement in the corticomedullary phase with peak enhancement that is significantly greater than that of other solid renal cortical tumors [10,11,12,13,14,15,16,17]. However, there are currently no published imaging biomarkers to predict the expression of phosphatase and tension homolog (PTEN) in clear cell RCCs. PTEN is a prognostically significant molecular marker in RCC. Lack of PTEN expression in RCC has been associated with worsened overall survival, a more advanced Fuhrman grade, and a greater likelihood of lymph node metastasis [18]. The purpose of this study was to investigate whether multiphasic MDCT enhancement can help to identify PTEN expression in clear cell RCCs.

Materials and methods

Patients

With Institutional Review Board approval for this Health Insurance Portability and Accountability Act-compliant retrospective study and a waiver of informed consent, we queried our institution’s pathology database to derive all histologically proven cases of clear cell RCC from December 2001 to August 2013. Of the approximately 1900 histologically proven cases of clear cell RCC obtained in this query, approximately 1770 lesions were excluded due to a lack of a preoperative dedicated four-phase renal mass protocol MDCT performed at our institution. An additional 31 cases were excluded due to a lack of availability of tissue specimens for immunohistochemical staining. This resulted in a cohort of 103 clear cell RCCs, representing 98 unique patients. Five patients had two lesions each. The mean time from multiphasic MDCT imaging to nephrectomy or biopsy was 66 days. The mean age at the time of nephrectomy or biopsy was 62 years, as shown in Table 1. Histopathology was obtained from partial nephrectomy [45 specimens (44%)], radical nephrectomy [31 specimens (30%)], and percutaneous 20-gauge core needle biopsy [27 specimens (26%)]. For tumors that were diagnosed at partial or radical nephrectomy, the T stage of each lesion was obtained by reviewing pathology reports for the resected specimens. For tumors that were diagnosed at percutaneous core needle biopsy, the T stage of each lesion was approximated based upon preoperative MDCT.

Table 1 Characteristics of clear cell RCCs
Full size table

MDCT examination

All CT examinations were performed at our institution using multidetector row CT [primarily 64 (2005–present) or 16-detector (2003–2008) row helical scanners (Sensation 64 or Sensation 16, Siemens Medical Solutions, Erlangen, Germany), or 4-detector (2000–2003) scanners (Lightspeed Qx/i, GE Medical Systems, Milwaukee, WI)]. Images were obtained with the patient breath holding with the following parameters: 120 kVP, variable mA, section thickness interval of 3 mm. A 4-phase renal mass protocol MDCT was performed for each patient. Each patient first underwent an unenhanced scan of the abdomen from the top of the diaphragm to the iliac crest. Then, following the injection of 100–150 mL of non-ionic intravenous iohexol based upon patient weight (Omnipaque 350, GE Healthcare, Milwaukee, WI) with a power injector at 3–4 mL per second, a bolus tracking algorithm (CareBolus, Siemens Medical Solutions; SmartPrep, GE Medical Systems) was utilized. For bolus tracking, a region of interest was placed in the thoracoabdominal aorta junction with a trigger set at 150 HU followed by imaging in the corticomedullary (40 s after the threshold is reached), nephrographic (90 s after the threshold is reached), and excretory (8 min after the threshold is reached) phases. Images were then sent to our picture archiving and communications system (Centricity, GE Medical Systems, Milwaukee, WI) for interpretation.

CT image analyses

Each lesion was segmented in its entirety on axial images in all phases using proprietary non-commercial Food and Drug Administration (FDA) 510(k) cleared software to create a three-dimensional contour. Lesion contours were drawn slightly internal to the lesion’s outer margin preclude partial volume averaging artifact [12, 19,20,21,22]. Each lesion contour was reviewed by two abdominal fellowship-trained genitourinary radiologists with 10 (MD) and 19 years (SSR) of experience, who were blinded to all clinical, histopathologic, and immunohistochemical findings. If there was any disagreement between the two readers regarding the appropriate placement of a contour, the case was discussed by both readers until a consensus was reached. A computer-assisted detection (CAD) algorithm utilized each whole lesion three-dimensional contour to select a 0.5-cm-diameter circular region of maximal attenuation within each lesion in each of the four phases. Pixels with attenuation > 300 Hounsfield units (HU) were excluded to exclude calcifications. Each CAD-selected region of maximal attenuation was reviewed by both readers for appropriate placement. A similar small region of interest approach to quantifying lesion attenuation has been utilized in prior studies examining multiphasic MDCT enhancement as a means of differentiating clear cell RCC from papillary RCC, chromophobe RCC, oncocytoma, and lipid-poor angiomyolipoma [12], identifying prognostically significant cytogenetic abnormalities in clear cell RCC (gain of chromosome 12, gain of chromosome 20, loss of the Y chromosome, and loss of chromosome 8p) [21, 23,24,25], identifying the level of carbonic anhydrase-IX expression in clear cell RCC [26], and differentiating type 1 from type 2 papillary RCC [20]. A similar approach has also been used to investigate multiphasic MR enhancement as a means of differentiating clear cell RCC from papillary RCC, chromophobe RCC, and oncocytoma [22]. For each lesion, a 0.5-cm-diameter region of interest was manually placed on uninvolved renal cortex in each phase, and these regions of interest were also reviewed by both readers for appropriate placement. In each phase, the relative attenuation (compared to uninvolved renal cortex) of each lesion was calculated as [(Maximal lesion attenuation − cortex attenuation)/cortex attenuation] × 100. Maximal lesion attenuation refers to the mean attenuation of the 0.5-cm circular region of maximal lesion attenuation. Cortex attenuation refers to the mean attenuation of the 0.5-cm circular region of interest manually placed on uninvolved renal cortex. Lesion size was determined by identifying the maximal lesion diameter on axial images.

Immunohistochemistry for PTEN

Surgical pathology specimens obtained from nephrectomy or percutaneous core needle biopsy were stained with a mouse monoclonal anti-PTEN antibody CM 278 (Biocare Medical, Pacheco, CA). Immunohistochemical staining was assessed by a fellowship-trained genitourinary pathologist (AS) to determine the presence of positively staining cells in each specimen.

Statistical analyses

To compare the maximal and relative attenuation values between PTEN-positive and PTEN-negative clear cell RCCs, t tests were performed using the hierarchical testing of the fixed sequence procedure by the order of contrast-enhanced phases (corticomedullary, nephrographic, and excretory). A multiple test procedure was controlled using a family-wise error rate at 5% [27]. The corticomedullary phase, which is the first contrast-enhanced phase acquired in our MDCT protocol, was placed first in the hierarchical testing as prior studies have shown that this phase is most important in differentiating clear cell RCC from other common RCC subtypes (papillary and chromophobe) and oncocytoma [10,11,12, 22] and in identifying prognostically significant chromosomal abnormalities in clear cell RCC [23]. Logistic regression models were constructed to evaluate the performance of attenuation parameters in identifying PTEN expression, after controlling for lesion size and stage. To account for potential clustering effects, the data were also analyzed while including only one lesion per patient in each group. For patients with multiple lesions, the representative lesion included was selected using a random number generator. Analyses were performed using SPSS 23 for Windows (IBM Corp, Armonk, NY).

Results

Lesions

Of the 103 clear cell RCCs in our study, immunohistochemical staining revealed that 19 lesions (18%) were PTEN positive and 84 lesions (82%) were PTEN negative. Baseline characteristics for lesions in each group are provided in Table 1. There was no significant difference in mean lesion size between PTEN-positive and PTEN-negative clear cell RCCs (4.2 cm vs. 4.0 cm, p = 0.85). The majority of PTEN-positive and PTEN-negative clear cell RCCs (68% in the PTEN-positive cohort and 76% in the PTEN-negative cohort) were either stage T1a or T1b at the time of diagnosis.

Multiphasic attenuation parameters

Both PTEN-positive and PTEN-negative clear cell RCCs exhibited peak maximal attenuation in the corticomedullary phase (Fig. 1). There was no significant difference in the maximal attenuation of PTEN-positive and PTEN-negative clear cell RCCs in the corticomedullary phase (Table 2); no further testing (in the nephrographic or excretory phases) was reported by the fixed sequence procedure. When including only one lesion per group per patient, relative corticomedullary phase attenuation (compared to uninvolved renal cortex) was significantly greater for PTEN-negative clear cell RCCs in comparison to PTEN-positive clear cell RCCs (35.5 vs. 9.5, p = 0.02, Table 3). However, no significant differences were found in nephrographic phase (p = 0.98), and no further testing was reported by the fixed sequence procedure. When including all lesions in the study cohort, the relative corticomedullary phase attenuation (compared to uninvolved renal cortex) was significantly greater for PTEN-negative clear cell RCCs in comparison to PTEN-positive clear cell RCCs (33.7 vs. 9.5, p = 0.03, Fig. 2, Table 4). No significant differences were found in nephrographic phase (p = 0.93), and no further testing was reported by the fixed sequence procedure. CT images of typical lesions in the PTEN-positive and PTEN-negative subcohorts are presented in Fig. 3. Immunohistochemically stained slides of typical lesions in the PTEN-positive and PTEN-negative subcohorts are presented in Fig. 4.

Fig. 1
figure 1

Maximal attenuation of PTEN-positive and PTEN-negative clear cell RCCs on multiphasic MDCT. Error bars represent 95% confidence intervals. For PTEN-positive clear cell RCCs, n = 19. For PTEN-negative clear cell RCCs, n = 84

Full size image
Table 2 Attenuation of clear cell RCCs on the basis of PTEN expression
Full size table
Table 3 Relative attenuation of clear cell RCCs on the basis of PTEN expression, including only one lesion per group per patient
Full size table
Fig. 2
figure 2

Relative attenuation of PTEN-positive and PTEN-negative clear cell RCCs on multiphasic MDCT. Error bars represent 95% confidence intervals. For PTEN-positive clear cell RCCs, n = 19. For the PTEN-negative clear cell RCCs, n = 84

Full size image
Table 4 Relative attenuation of clear cell RCCs on the basis of PTEN expression, including all lesions in the study cohort
Full size table
Fig. 3
figure 3

Corticomedullary phase axial MDCT images of a PTEN-positive clear cell RCC in a 71-year-old woman and a PTEN-negative clear cell RCC in a 67-year-old woman. Representative CAD-selected regions of maximal attenuation are shown in blue. A PTEN-positive clear cell RCC with a maximal corticomedullary phase attenuation of 141 HU. B PTEN-negative clear cell RCC with a maximal corticomedullary attenuation of 211 HU

Full size image
Fig. 4
figure 4

Immunohistochemically stained slides of a PTEN-positive clear cell RCC (A) and a PTEN-negative clear cell RCC (B)

Full size image

Logistic regression modeling

A logistic regression model was constructed to evaluate the performance of relative corticomedullary phase attenuation in identifying PTEN expression in clear cell RCCs, after controlling for lesion size and lesion stage. After controlling for these variables, relative corticomedullary phase attenuation had an odds ratio of 0.985 (p = 0.06). The model (with a classification cut-off of 0.5) had an accuracy of 84% (86/103), specificity of 100% (84/84), sensitivity of 11% (2/19), positive predictive value of 100% (2/2), and negative predictive value of 83% (84/101) in differentiating PTEN-positive from PTEN-negative clear cell RCCs.

Discussion

Consistent with prior published studies evaluating the multiphasic MDCT enhancement of renal tumors, clear cell RCCs in our study cohort typically exhibited peak enhancement in the corticomedullary phase [10,11,12,13,14,15,16,17]. In this current study, we found that relative corticomedullary phase attenuation (compared to uninvolved cortex) may help to identify PTEN expression in clear cell RCCs. PTEN-positive clear cell RCCs had a significantly lower relative corticomedullary phase attenuation than PTEN-negative clear cell RCCs. In a logistic regression model that controlled for lesion size and lesion stage, relative corticomedullary phase attenuation had an accuracy of 84% (86/103), specificity of 100% (84/84), sensitivity of 11% (2/19), positive predictive value of 100% (2/2), and negative predictive value of 83% (84/101) in differentiating PTEN-positive from PTEN-negative clear cell RCCs.

PTEN is a prognostically significant molecular marker in RCC. It is a tumor suppressor that downregulates several important signal transduction pathways, including the hypoxia-inducible factor 1-α (HIF-1α), PI3 kinase, and mitogen-activated protein (MAP) kinase pathways. Lack of PTEN expression in RCC has been associated with worsened clinical outcomes and worsened overall survival rates [18]. PTEN-negative RCCs typically have a more advanced Fuhrman grade and a greater likelihood of lymph mode metastasis than PTEN-positive RCCs [18]. Additionally, prior studies have shown that the loss of PTEN expression is negatively correlated with clinical response to anti-vascular endothelial growth factor (VEGF) and anti-epidermal growth factor (EGF) therapies [28]. Tumor resistance to anti-VEGF therapies is thought to be due to a switch from VEGF-dependent or EGF-dependent angiogenesis to HIF-1α-dependent and PI3 kinase-dependent angiogenesis [28]. Furthermore, given the importance of PTEN in a variety of signal transduction pathways, microRNA therapies that target PTEN have been the focus of recent investigations. Liu et al. found that PTEN is a direct target of miR-193a-3p and that downregulation of miR-193a-3p inhibited cellular proliferation in RCC cell lines [29].

In our cohort, 82% (84/103) of the clear cell RCCs were PTEN negative. This percentage of PTEN-negative RCCs is higher than in other published studies. In the study by Zhu et al., 59% (51/87) of RCCs were PTEN negative [18]. PTEN is a downregulator of the HIF-1α and PI3K signal transduction pathways. The PI3K signaling cascade activates serine/threonine kinase AKT. Activated AKT phosphorylates multiple substrates that regulate numerous cellular functions, including cellular growth, metabolism, and tumorigenesis [30]. Data from the Cancer Genome Atlas have shown that alterations in the PI3K signaling pathway are frequently encountered in clear cell RCC [30]. Additionally, prior studies have shown that the loss of PTEN expression is negatively correlated with clinical response to anti-VEGF and anti-EGF therapies [28]. Because of the prognostic significance of PTEN expression, a simple, non-invasive means of identifying PTEN expression in clear cell RCCs may be helpful in assessing patient prognosis and may potentially guide further management, if validated prospectively.

Our study has several potential limitations. First, while all lesions in our study were evaluated with a dedicated four-phase renal mass protocol MDCT, due to the retrospective nature of this study, we could not account for variability of absolute attenuation within the same phase related to differences among scanners. However, our institution’s four-phase renal mass MDCT protocol was designed to standardize scanning parameters as much as possible. Second, this was a preliminary, discovery-phase study investigating the relationship between PTEN expression and multiphasic MDCT attenuation. Our findings should be validated in an independent testing cohort. Although we do not have a definitive mechanistic explanation for the relationship between PTEN expression and relative corticomedullary phase attenuation, we suspect that this may be related to differences in angiogenesis. PTEN is an important modulator of angiogenesis through its inhibitory effects on the HIF-1α signal transduction pathway, which promotes angiogenesis [28]. Third, clear cell RCCs commonly exhibit intratumoral heterogeneity with regard to genetic mutations and resultant protein expression [31]. Because of this heterogeneity, the degree of PTEN expression may vary at different sites within a particular tumor. In addition, the CAD-selected region of maximal attenuation may not necessarily correspond to the portion of the tumor that was immunohistochemically stained for PTEN expression.

Fourth, T stage assessment differed between lesions that were diagnosed at partial or radical nephrectomy and those that were diagnosed at percutaneous core needle biopsy. The T stage of lesions diagnosed at partial or radical nephrectomy was obtained by reviewing pathology reports for the resected specimens, while the T stage of lesions diagnosed at percutaneous core needle biopsy was approximated based upon preoperative MDCT. Fifth, in a logistic regression model that controlled for lesion size and lesion stage, relative corticomedullary phase attenuation had an accuracy of 84% (86/103) in identifying PTEN expression in clear cell RCCs, which is not yet adequate to definitively guide further treatment. However, our findings suggest that there may be a consistent relationship between PTEN expression in clear cell RCCs and multiphasic MDCT enhancement, if validated in a large prospective trial.

To our knowledge, our study is the first proof of concept study to demonstrate an MDCT-based difference between PTEN-positive and PTEN-negative clear cell RCCs. Because of the prognostic significance of PTEN expression, a non-invasive means of identifying PTEN expression preoperatively may help to guide clinical decision making and patient counseling. Furthermore, with baseline multiphasic MDCT imaging prior to surgery or ablation, new lesions may be non-invasively evaluated to determine if their enhancement profiles are similar to previously treated lesions. Future work is needed to determine whether our findings can be replicated in a prospective cohort and to determine whether our findings can be replicated with gadolinium-enhanced magnetic resonance imaging and contrast-enhanced ultrasound imaging. Future work should also evaluate other imaging techniques, such as positron emission tomography, magnetic resonance diffusion-weighted imaging, and magnetic resonance elastography, to identify additional imaging biomarkers of PTEN expression.