Lymphedema after treatment for endometrial cancer − A review of prevalence and risk factors

The reported prevalence of LLL after treatment for EC ranged from 0% [] up to 50% [] (Table 1). The design of the majority of the studies giving information about the prevalence of LLL was retrospective or descriptive cross-sectional. Only two of the published articles were randomized controlled trials (RCT) but none of these were designed to evaluate LLL as the primary outcome measure. The two RCTs compared the use of LA versus no LA in surgical treatment for EC, with the primary outcomes being overall − and recurrence free survival [, ]. In these studies the patients who had LA showed a higher prevalence of LLL, 13% and 3.5%, respectively, compared with 1.6% and 0.3%, respectively, for those without LA [, ]. However, both studies lacked baseline information concerning leg swelling, and the method of determining the occurrence of LLL was not clearly described. One recently published prospective cohort pilot study aimed to determine the incidence of LLL using objective measurements of the lower limbs, and found that 13% of patients had LLL within two years of surgery []. The criteria for the diagnosis of lymphedema in that study was an increase in leg circumference of 20%.

Many of the retrospective studies were designed to compare the surgery ± LA, extent of the LA, different surgical techniques or modes of surgery in the treatment of EC. The prevalence of LLL varied considerably between these studies, from 0% [] to 50% []. Besides being retrospective the common denominators for all these studies were that they might be characterized by either having no baseline data concerning leg swelling, lacked objective measures or were based on incompletely described or not validated subjective measurements. In some studies only cases with moderate to severe lymphedema or cases that complained of leg swelling and received treatment were classified as having LLL [, , ]. Consequently the true prevalence of LLL was not obtained in these studies.

In those studies where objective measurements of the LLL had been conducted the prevalence of LLL seemed to be higher than in those with subjective measures based on patient record information only. The prevalence varied from 1.2% to 28% [, ] in the studies with subjective measurements with one exception that showed a considerably higher prevalence (50%) []. In the studies with objective measurements the prevalence varied from 7.0% to 38% [, ].

Several methods to evaluate and categorize lymphedema, e.g. American Physical Therapy Association lymphedema rating system (APTA), the Late Effects of Normal Tissue/Subjective, Objective, Management, Analytic components three item system (LENT/SOMA) and the Common Toxicity Criteria with the lymphedema criteria (CTC v2.0) have been presented but are not widely used in studies of cancer treatment. A consensus statement was presented by Cheville et al. in 2003 that recommended the use of the Common Terminology Criteria for Adverse Events (CTCAE v 3.0) in clinical trials []. Although the system used for the classification of lymphedema is not universally accepted, the International Society of Lymphology (ISL) issued an updated consensus statement in 2009 and 2013 regarding diagnosis and treatment of lymphedema []. The ISL describes four stages of lymphedema. The severity of changes within a stage may be determined by measuring the volume differences. The assessment of LLL varies widely in the published articles of EC. Still there seems to be no universally used standardization for how to measure or report LLL in scientific contexts. Table 2 summarizes the mode of assessment − subjective or objective measures − used in the published articles reviewed in this study. Only one recently published prospective study that used objective measurements of the limbs at baseline, i.e. before the cancer treatment, was identified []. In two Japanese studies circumferential measurements of the limbs were performed but only after the surgical treatment [, ]. In many of the studies the assessment of LLL was based on information retrieved from the medical records or from patient reported symptoms in the survey studies. The ISL 4-grade staging of lymphedema [] has been used by some authors [, , , , ], whereas others have used the National Cancer Institute Common Toxicity Criteria (NCI-CTC v2 []) [, , , ] or the National Lymphedema Network [] to grade the severity of lymphedema.

For the subjective scoring of LLL symptoms questionnaires have been developed in order to establish an easy assessment based on patient-self-reported symptoms (European Organization for Research and Treatment of Cancer Quality of Life Questionnaire − Endometrial Cancer Module (EORTC-QLQ-EN24) and) Self-Report Lower-Extremity Lymphedema Screening Questionnaire in Women) [, ]. Such a method could easily be incorporated into the clinical care setting as a simple, feasible and time-efficient screening/triaging method to identify women at a high risk of developing lymphedema. In the review of the literature, the total prevalence of LLL based on symptom specific questionnaires varied from 8.2% [] to 47% []. However, in the study by Yost et al., only 17% of those who according to the questionnaire, were considered to have LLL, had previously been clinically diagnosed with LLL. This may indicate a substantial underestimation of the occurrence of LLL by simple clinical estimation [] or an overestimation when assessing using a questionnaire only.

Risk factors evaluated in the studies are depicted in Table 1. LA per se has been shown to be an independent risk factor for LLL in several studies [, , , ]. LA is not a unanimous surgical concept but differs considerably in extent of anatomical areas evacuated for lymph nodes and the number of lymph nodes removed.

The extent of LA varies from sampling of a few lymph nodes to removal of median 10–30 pelvic and/or para-aortic lymph nodes [, , , , ]. In other studies the median number of removed lymph nodes is significantly higher, up to over 80 [, , , ]. The number of removed lymph nodes seems to be of importance for developing LLL with an increasing risk for LLL when a larger number of lymph nodes are removed [, , , , , , , , , ]. Interestingly, and with no obvious reasonable pathophysiological explanation, Hareyama et al. found a significantly lower risk of LLL in women when the number of removed lymph nodes exceeded 70 [].

The site of the lymph nodes removed may also increase the risk for LLL in gynecologic cancer. Removal of the circumflex iliac lymph nodes to the level of the distal external iliac nodes (the so called CINDEIN, also named CILN (circumflex iliac lymph nodes) or Cloquet’s node) gave a significantly higher incidence of LLL [, , ]. The topographic extent of the LA seems to matter less. To perform both pelvic and para-aortal LA has not been shown to lead to more LLL than LA confined to the pelvic area.

The occurrence of lymph node metastases in the removed lymph nodes was shown by Mitra et al. to be a risk factor for LLL [], but the results are contradictory [, ].

Pelvic radiation therapy (RT) given alone or as adjuvant therapy after surgery has a varying impact on the development of LLL. In several studies RT has been shown to significantly increase the risk for LLL [, , , , , , , , , ]. However, there are some studies that fail to demonstrate an association between RT and LLL [, , ]. Adjuvant chemotherapy, either given alone or in combination with RT, does not seem to have the same impact but in a few studies has been shown to increase the risk for LLL [, ].

Overweight and obesity are associated with increased risk for EC. Therefore the populations of EC survivors may also be expected to have higher BMI than the average population. In three of the descriptive studies where the participants answered a questionnaire concerning LLL, higher BMI was significantly associated with more LLL symptoms [, , , ] while in other studies such an association between BMI and risk for LLL was not demonstrated [, , , , ].

Higher levels of physical activity have been associated with a significantly reduced risk for LLL. In the study by Brown et al. [] a higher level of physical activity was a preventive factor for LLL but only among women with BMI <30 kg/m².

Age has also been associated with the development of LLL in gynecological cancer but the results are contradictory. Deura et al. [] found age over 55 years to be an independent risk factor for LLL. Ferrandina et al. showed that women over 60 years had a higher risk for LLL but only in a univariate analysis []. Several studies have shown no such association between age and risk for development of LLL [, , , , , ]. Thus, it remains unclear whether BMI or age truly are risk factors for LLL in EC patients.

In the large retrospective review by Abu-Rustum et al. including 1289 patients, 5.7% reported swelling of the legs due to other medical conditions []. Survivors of EC as a population are likely to be older and to have higher BMI, resulting in more co-morbidity including being at risk for other medical conditions which may give symptomatic swelling of the legs. Yost et al. [] found that congestive heart failure was an independent risk factor for LLL. The diagnosis was based on a self-reported screening questionnaire. Whether this finding reflected a true increased risk of LLL or simply a swelling of the legs on the basis of cardio-vascular disease was not evaluated.

A summary of the studies on mode of surgery for EC and LLL is shown in Table 3. Minimal invasive surgery is considered theoretically to decrease the risk of developing LLL. However, this remains to be proven. In a prospective study, women with EC were randomized to laparoscopic surgery (LPS) or laparotomy (LPT). In the LPT group, 3.4% developed LLL compared with none in the LPS group []. However, the study gave no information on how the LLL diagnosis was obtained. Several retrospective studies have compared LPS to LPT in EC and in some cases also robot-assisted LPS. The reported incidence of LLL vary from 0.9% [] to 1.6% [] with a higher incidence after LPS in two of the studies [, ]. Another study showed no difference between the groups at all []. Two studies compared LPT, LPS and robot-assisted LPS [, ]. However, only one patient were diagnosed with LLL in each of these two studies; one in the LPT group in [], and one in the robot-assisted LPS group in []. When comparing conventional LPS with robot-assisted LPS no difference in prevalence of LLL was found, 2% and 3%, respectively []. In four studies with only robotic surgery comprising pelvic and/or para-aortic LA [, , , ], the LLL incidence was reported to be 5.2% to 21%, with a lower incidence, although not statistically significant, when the operation had been converted to LPT []. In all of these studies the diagnosis of LLL was based on information from the medical records, with no systematical measurements performed except in the study by Geppert et al. [] who used the NCI-CTC classification prospectively on those with symptomatic LLL but without baseline evaluation.