Introduction

Cancer incidence and improved cancer survivorship have been increasing globally over the last few decades primarily due to population growth, aging populations, advances in cancer detection methods and changes in lifestyle and risk factors [1]. With an estimated 20 million new cancer cases reported annually worldwide, comes a plethora of long-term sequelae related to the cancers and their respective therapies, ranging from cardiotoxicity, functional impairments, cognitive issues, psychosocial and physical late and long-term effects, including cancer-related lymphedema (CRL) [2,3,4].

Cancer-related lymphedema is a chronic inflammatory process in the interstitial space due to reduced lymph transport capacity from damage to the lymph vessels, nodes or by direct tumor involvement [5,6,7]. CRL is progressive and may become a very debilitating condition for many cancer survivors, affecting the person physiologically, physically, and emotionally [8]. Because CRL requires lifelong management, cancer survivors often describe it as one of the most significant sequela of cancer treatment [2, 9, 10]. In the case of active cancers, malignant lymphedema may arise from infiltration, obstruction, or compression of lymphatic vessels and/or lymph nodes by the direct action of the tumor [11].

The impact of lymphedema on quality of life (QOL) can vary depending on the severity of the condition, its management, and the individual’s emotional, psychological, and physical well-being. Some of the extensively documented, significant effects of lymphedema on an individual’s QOL include physical discomfort, altered body image, reduced self-esteem, functional impairment, psychological distress, social barriers, occupational stressors, marginalization. In addition, patients may experience financial burdens because of costs associated with intensive treatment phases and lifelong expenses for compression garments [8, 12, 13]. Beyond these considerable individual impacts, there exists an economic strain on payers, healthcare systems, and society due to lymphedema-related hospitalizations for complications such as cellulitis and sepsis, as well as interruptions in employment [14,15,16].

Current estimates suggest that approximately 10 million individuals in the United States [17] and around 1 million in Canada [7] are affected by lymphedema. The etiology of lymphatic dysfunction could be primary, due to malformation of the lymphatic structures (vessels or nodes), or secondary, due to chronic lymphatic system overload (e.g., chronic venous insufficiency, obesity, CRL, trauma). Despite various causes of secondary lymphedema, with CRL being the most extensively documented, accurate estimates remain elusive due to inadequate documentation.

The reported incidence of CRL varies widely depending on cancer type, associated treatments and individual variability. Several factors contribute to the complexity of estimating CRL incidence, including the absence of a standardized international definition of lymphedema (e.g., definition based on varied clinical volume differences, self-report of swelling), variability in diagnostic assessment methods and criteria (e.g., circumference measurements—anatomical landmarks, every 10 cm; perometry), lack of pre- and post-operative screening protocols, and inconsistencies in timing and duration of follow-up assessments.

To varying degrees, CRL may affect many cancer survivors who have undergone node dissection surgery, chemotherapy, and/or radiotherapy. In this review, we expand upon the groundwork laid by Cormier et al. (2010) [9] and Shaitelman et al. (2015) [18] to provide updated insights into the incidence of CRL across a wide range of cancers. Estimating CRL incidence is crucial to inform care access and resources allocation needs to best support patients affected by this lifelong condition.

Methods

Search strategy

A literature review was performed in two phases. The first phase was conducted by a research librarian (DPF) who searched in the following four databases: MEDLINE, Embase, Web of Science Core Collection, and Cochrane Library. Publications dated between 2015 and 2023 were included, to gather entries released since previously published reviews on the topic [9, 18]. Using controlled vocabularies adapted to each respective database (Mesh; Emtree), search strategies were further refined, using the main keywords “lymphedema,” “cancer,” and “prospective studies.” We excluded publications on non-cancer-related lymphedema, animal studies, letters, small case series, case reports (n < 50), and those written in a language other than English. A search hedge was adapted for the prospective studies entry material [19]. The initial search yielded 3756 entries (Fig. 1).

Fig. 1
figure 1

Flow diagram of search results

Full size image

In the second phase of the review, initial results were uploaded to Covidence, a web-based software platform that streamlines the production of reviews [20]. After we excluded duplications (n = 1700), titles and abstracts of 2058 entries were screened by two reviewers (MEL and SS), yielding a total of 146 studies. Two independent reviewers (MEL and DT) then performed a detailed review of these studies in accordance with predefined inclusion criteria. We required prospective assessment of CRL as a primary or secondary outcome and a minimum sample size of 50 patients. In keeping with the main goal of estimating CRL incidence, we also included retrospective reviews presenting prospective collection of lymphedema-related data. We therefore retained 68 studies.

Quality control

Of the 13 JBI critical tools available, checklists for case controls [21], prevalence studies [22], randomized controlled trials [23], and systemic reviews [24] were utilized to assess these 68 studies. Each checklist concludes with a decision as to whether to include, exclude it, or seek additional information. Two of four reviewers (MEL, MI, AT, DT) independently evaluated each study. In case of disagreement, a third reviewer (MEL or MI) acted as an arbitrator. This process yielded a final total of 48 studies.

Statistical analysis

The choice of methods for summary and subgroup analysis was limited by the heterogeneity of the data due to the absence of a standardized clinical lymphedema definition, variations in measurement techniques and length of follow-up. When feasible, we conducted subgroup analyses based on cancer type. Predefined review characteristics for analysis included sample size, type of lymphedema assessment (objective and subjective), measurement methods, and length of follow-up. Radiation therapy and lymph node dissection variables were also considered, when data were available.

Results

The search of the four medical indices yielded a total of 48 articles (33 prospective studies and 15 systematic reviews), providing a total sample size of 234,079 cancer survivors (Fig. 1). Our search encompassed papers reporting on CRL according to any cancer type. However, only four categories of cancer were included breast, gynecological and urological, head and neck (H&N) and melanoma. A fifth category represented preventive lymphatic surgery aiming to reduced CRL incidence including lymphovenous anastomosis (LVA) and axillary reverse mapping (ARM).

Breast cancer studies were the most numerous (50%, n = 24) and provided 88.6% of the total study population. Gynecological and urological cancers represented 27.1% (n = 13) of the included articles, and 6.1% of the population. This was followed by preventive lymphatic surgery (12.5%, n = 6), mainly focussing on breast cancer patients, representing 4.3% of the total population. Melanoma and H&N cancers represented the lowest number of included articles with 6.3 and 4.2%, respectively, corresponding to 0.5 and 0.2% of the total population.

A total of nine clinical definitions were identified, whereby 43.8% of the studies used a difference of greater than 5–10% between the limbs to define lymphedema (Table 1). Additionally, some studies also used more than one definition when various methods of diagnosis were used. About a quarter of the studies (13 studies, 27%) did not specify their clinical definition of lymphedema.

Table 1 Lymphedema clinical definitions
Full size table

Our review yielded a plethora of diagnostic methods, grouped into seven different methods, whereby most studies (30 studies, 62.5%) used circumferences and self-reported (16 studies, 33.3%) methods (Table 2). Several studies used more than one method of diagnosis, while two studies did not report diagnostic methods.

Table 2 Diagnostic methods for lymphedema
Full size table

Subsequently, while extracting data, we were able to identify several trends regarding risk factors that might influence CRL incidence. Our findings suggest that the following could increase the risk of developing CRL: extensive surgery (e.g., tumor and/or lymph node dissection), increasing age, having received adjuvant therapy (both chemotherapy and radiation therapy), body mass index (BMI) greater than 25 or 30 at surgery and increasing post-treatment, insufficient physical activity levels, and post-surgical complications (e.g., lymphocele, wound infection, seroma). Conversely, education (preoperative and general about lymphedema), prospective surveillance with early identification of signs and symptoms, lymphedema risk reduction practices, weight management or weight reduction, scar tissue release and omission of axillary dissection, show a trend toward decreasing the risk of CRL.

Breast cancer

A total of 15 prospective studies reporting on the incidence of breast cancer-related lymphedema (BCRL), representing nearly 15,000 patients, were identified (Table 3) [25,26,27,28,29,30,31,32,33,34,35,36,37,38,39]. Incidence findings were reported using either objective measures (limb circumference, Bio-Impedance Spectroscopy (BIS), perometry, Indocyanine Green (ICG)), or by self-report. Length of follow-up varied between 6 and 120 months.

Table 3 Studies reporting on BCRL incidence
Full size table

For objective measures, BCRL incidence ranged from 2 to 60.3%. While the majority of studies (9 studies) utilized circumferential measurements to assess for lymphedema, variations in how lymphedema was defined were noted. Lowest incidence was reported with perometry and ICG at 24 months. For self-report, BCRL incidence ranged from 2.1 to 74.3%. Survivors treated with axillary lymph node dissection (ALND), chemotherapy and/or regional lymph node radiation reported a higher incidence of BCRL, varying between 2.9% (perometry and ICG) and 60.3% (circumference). Table 3 further highlights the heterogeneity of the findings reported for the breast cancer population.

Additionally, nine systematic reviews reporting on BCRL incidence met our inclusion criteria, encompassing a total of 283 articles and representing over 192,000 BCRL patients (Table 4) [40,41,42,43,44,45,46,47,48]. Some systematic reviews provided pooled lymphedema estimates, varying between 4 and 23.6%, and/or a variation in lymphedema incidence ranging from 0% to 63.4%.

Table 4 Systematic reviews reporting on BCRL incidence
Full size table

Some systematic reviews presented lymphedema incidence based on the measurement technique used. Water displacement tends to report the lowest incidence (0–22.2%), followed by BIS (3.1–31.9%), self-report (0–54.0%), perometry (7.2–54.0%), and circumferences measurements (1–63.4%). As observed in prospective studies, ALND showed a higher incidence in the systematic reviews varying from 6 to 24.6%, compared to 3.7% to 7.5% for SLNB.

Gynecological and urological cancers

Eleven studies, with over 3200 patients, representing cervical, endometrial, ovarian, and vulvar malignancies were included (Table 5) [49,50,51,52,53,54,55,56,57,58,59]. The length of follow-up varied between 6 and 120 months, with the highest incidence of 45.1% self-reported by patients at 24 months. Incidence varied among the different types of gynecological cancers, with all of them reporting a lower incidence when neither lymph node dissection nor sentinel lymph node biopsy (SLNB) were performed. The highest incidence was reported by women treated for vulvar (66.7%), followed by cervical (51.7%), endometrial (44.4%), and ovarian (40.4%) cancer. Most studies assessing for lower limb lymphedema (LLL) post-gynecological cancers used circumferential measurements as an objective measure (Table 5).

Table 5 Studies reporting on gynecological CRL incidence
Full size table

One systematic review on prostate cancer [60] and one on vulvar cancer [61] were included with a total sample size of 11,758 patients (Table 6). For prostate cancer, Clinckaert et al. (2022) [60] reported LLL varying from 0 to 29%, and genital lymphedema varying from 0 to 22%, respectively. A higher incidence was found in those who underwent pelvic lymph node dissection (PLND) and radiotherapy. Huang et al. (2017) [61] provided a pooled LLL estimate of 28.8% in vulvar malignancies, with the highest incidence in cross-sectional studies (49.2%) or randomized controlled trials (45.1%).

Table 6 Systematic reviews reporting on gynecological and urological CRL incidence
Full size table

Head and neck cancers

A total of 380 participants, distributed in two studies were included (Table 7) [62, 63]. Participants of both studies received similar interventions. The incidence of H&N CRL tends to be higher in the early phase post-treatment, varying from 80 to 90.1%, while decreasing over time to 70.6–82.3%. Ridner et al. (2016) [62] reported that external lymphedema tends to vary between 81.9 and 90.1%, internal lymphedema between 80.4 and 89.4%, and a combination of both between 70.6 and 80.9%. Tribius et al. (2020) [63] reported an incidence of 80% between 3 and 6 months in advanced stage H&N cancer.

Table 7 Studies reporting on Head and Neck CRL incidence
Full size table

Melanoma cancers

The three melanoma studies all distinguished between upper and lower limb lymphedema (Table 8) [64,65,66]. The overall CRL incidence varied between 2 and 28.6%, with the lowest incidence attributed to upper limb or trunk melanoma treated with SLNB (1.0–18.4%), and the highest to lower limb or trunk melanoma treated with “total” lymph node dissection (TLND) (7.7–47.4%). Only one study [65] had participants self-reporting an incidence of 23.1% at a median time of 37 months after SLNB in both upper and lower limb CRL.

Table 8 Studies reporting on Melanoma CRL incidence
Full size table

Preventive surgery

Two prospective studies [67, 68] and four systematic reviews [69,70,71,72] on surgery aiming to prevent lymphedema were included, representing a total of 10,080 patients (Table 9). Preventive surgery included LVA or ARM procedures, in cases where full lymph node dissection was judged to be required. The reported CRL incidence was 16–28.5% and 5.2–23.4% for prospective studies and systematic reviews, respectively. Patients receiving the preventive procedure experienced a lower incidence of CRL compared to controls in both prospective studies and systematic reviews: 3–21% versus 19–42%, 2–18% versus 14.1–48.5%, respectively.

Table 9 Studies and systematic reviews reporting on preventive surgery CRL incidence
Full size table

Discussion

In terms of trends in the incidence of CRL, previous work from Cormier et al. (2010) [9] published an overall CRL incidence of 15% for all cancer site (melanoma, genitourinary, gynecological cancers, excluding BCRL). Shaitelman et al. (2015) [18] extended the work from Cormier et al. (2010) [9] and published in addition a pooled incidence of 6.3% for BCRL in patients who underwent SLNB, and 22.3% for BCRL in patients who underwent ALND. For the purpose of this updated review on incidence of CRL, given the heterogeneity of data pertaining to lymphedema definitions, diagnostic methods, and variations in length of follow-up, meta-analysis, and pooled incidence could not be conducted. Findings were therefore reported by cancer site to provide tentative inferences.

For BCRL post-ALND, Shaitelman et al. (2015) [18] reported objective incidences varying from 11 to 57%, compared with our findings of 3–60%. For SLNB alone, Shaitelman et al. (2015) [18] reported a rate of 0–23%, whereas our review narrowed the range to 1–12%, which might be attributed to the fact that most studies (4 studies, 57.1%) used perometry to assess BCRL. Our review was able to provide additional information in terms of a variation of 3–31% of BCRL incidence with radiation treatment, and 5–60% when chemotherapy is a modality of treatment. Therefore, the incidence of BCRL remains important, especially when combined with ALND, regional lymph node radiation and chemotherapy.

For gynecological cancers, previous reviewers reported an incidence rate of 0–73%, with the highest rate observed in vulvar cancer (0–73%), followed by cervical cancer (2–49%), and lastly for endometrial cancer (1%). Our review diverges significantly from these findings, revealing a notably higher rate for endometrial cancer (1–42%), along with a narrowed range for vulvar cancer (8–43%) and for cervical cancer (14–52%). The cumulative incidence rate for gynecological cancers also shows a reduced range, from 0 to 73% in previous reviews to 11–41% in our findings. Moreover, SLNB alone decreases the rate, with our results aligning with previous reviews showing a rate of 0–25%, compared to our 1–31%.

In terms of genitourinary cancer incidence, the two preceding reviews indicated a variation of 1–18% for prostate cancer, while our findings show a broader range of 0–29%. Additionally, those reviews covered penile and bladder cancer, reporting rates of 20–21% and 15–23%, respectively. Unfortunately, we encountered a lack of available studies to enable an update on the incidence for those two cancers.

For H&N cancers, our results differ markedly from the previously published reviews. Cornier et al. (2010) [9] and Shaitelman et al. (2015) [18] reported a range of 0–8%, whereas more recent studies suggest a rate of 80–90%. This concurs with our clinical experience.

For melanoma, there has been a significant reduction and narrowing of cumulative ranges. Previous studies indicated an overall range of 1–61%, whereas we report a range of 2–29%. While the ranges for lower extremity lymphedema following total lymph node dissection for melanoma have narrowed, they still remain substantial: 6–61% in previous reviews, compared to 8–47% in our findings. An unexpected finding is the substantial lymphedema rate after SLNB alone: 1–15% in previous reviews and 1–32% in our study.

Preventive lymphatic surgeries, such as LVA or ARM, performed at the same time of lymph node dissection are relatively recent innovations. Consequently, there was no previous comparison data on the incidence rates following these preventive procedures.

The existing data on lymphedema incidence remains primarily focused on the breast cancer population. A paucity of evidence persists for non-breast CRL, including gynecological (uterus, ovaries, cervix, or vulva), urological (prostate, bladder, urethral, kidney, testicular, and penile cancers), gastrointestinal (colorectal, anal, bile duct, pancreatic, gastric, and liver cancer), melanoma, H&N cancers, lymphoma, and sarcoma. Given that clinically we observe CRL in many of these patients, a collaborative effort is needed to capture its incidence in these populations that are underrepresented in research studies.

As concluded by Cormier et al. (2010) [9] and Shaitelman et al. (2015) [18], a significant challenge in lymphedema research lies in establishing consensus among researchers regarding diagnostic standards and measurement techniques. Researchers and clinicians use various objective tools and methodologies to diagnose lymphedema, such as clinical evaluation, circumferential measurements, bioimpedance, water displacement, perometry, imagery, and self-report measures. This diversity in diagnostic approaches contributes to a wide range of reported incidence rates, complicating comparisons between studies. Research settings using technology, such as BIS or perometry, may tend to report a lower lymphedema incidence. On the other hand, research settings using circumferential techniques for its ease and low cost of use, and more likely representative of clinical practice, are potentially reporting a higher incidence of lymphedema. Studies relying solely on one diagnostic method may underestimate the true incidence of lymphedema.

While objective tools are crucial for diagnosing lymphedema, self-reported symptoms such as swelling, sensation of heaviness, perceived limb size difference and discomfort play an equally vital role in initial diagnostic screening [73]. Healthcare professionals must attentively consider these patient self-reported symptoms, as they may facilitate early detection of lymphedema and lead to earlier management, reduced complications and financial burden, while improving a patient’s QOL.

Conclusion

Our findings revealed significant CRL incidence across several cancer types. Early identification of lymphedema signs and symptoms and prompt referral to a certified lymphedema therapist are crucial to prevent the myriad complications resulting from inadequate management of this chronic and progressive condition. As patients increasingly survive cancer treatment, clinicians bear responsibility to minimize the burden on patient’s QOL from the multitude of potential side effects that may arise. To facilitate future research, the international lymphedema community should make it a priority to converge on accurate, reproducible, accessible, cost-effective, reliable, and quantifiable diagnostic methods. Such standardization efforts would enhance research quality and mitigate discrepancies in reported incidence rates in the future.