Hysterectomy and kidney cancer risk: a meta-analysis.

In total, seven cohort studies [ 5 , 13 – 17 ] and nine case-control studies [ 18 – 26 ] were identified. A brief description of all studies is provided in Supplemental Table 1 . Three case-control studies [ 22 – 24 ] were excluded because they represented subsets of a large international study [ 21 ]. Thus, results from thirteen studies (7 cohort, 6 case-control) were included in the meta-analysis. A forest plot summarizing study-specific and summary associations between hysterectomy and kidney cancer is illustrated in Figure 1 . The SRR for hysterectomy and kidney cancer for all published studies combined was 1.29 (95% CI, 1.16 to 1.43), with no evidence of between-study heterogeneity ( P =0.11) or publication bias ( Supplemental Figure 1 ). The summary effect was slightly weaker, although still significant, for cohort findings (SRR, 1.26; 95% CI, 1.11 to 1.42; P heterogeneity =0.07) compared to case-control findings (1.37; 95% CI, 1.09 to 1.73; P heterogeneity =0.28). Publication bias was a concern for cohort studies ( P -Begg’s=0.051; P -Egger’s=0.02); yet, after removal of one influential study [ 13 ], publication bias was no longer observed and the SRR remained significantly elevated (1.21; 95% CI, 1.09 to 1.34). Furthermore, the summary effect did not materially change, as shown in Table 1 , when we conducted meta-analyses of risk estimates for hysterectomy without oophorectomy and for hysterectomy with adjustment for body mass index (BMI), smoking, and hypertension. The summary estimate also did not substantially change when assessing studies (N=4) that adjusted for education as well as established kidney cancer risk factors (SRR, 1.49; 95% CI, 1.24 to 1.78; P heterogeneity = 0.39; P -Begg’s= 0.17; P -Egger’s= 0.13). Similarly, results did not change for meta-analyses conducted by age at hysterectomy, years since hysterectomy, and geographic region ( Table 1 ).

In this meta-analysis we found hysterectomy was associated with a statistically significant elevated relative risk of developing kidney cancer, both from cohort and case-control findings. Positive findings reported in a number of previous epidemiological studies have been criticized for not accounting for kidney cancer risk factors that may confound results, for not assessing the impact of other co-gynecological procedures such as oophorectomy status, and for not evaluating time since hysterectomy to address issues related to detection bias [ 13 , 17 , 30 ]. However, our meta-analysis suggests that the hysterectomy association with kidney cancer is not affected by, or sensitive to, these factors. Our subgroup analyses show, regardless of study design, hysterectomy is positively associated with kidney cancer risk, irrespective of time since hysterectomy, study location, or adjustment of established kidney cancer risk factors. Significantly elevated estimates for subgroup analyses by age at hysterectomy and hysterectomy without an oophorectomy were significant only for cohort studies and cohort and case-control studies combined. The lack of significant findings for case-control studies is likely due to statistical power limitations given the small case numbers available for analysis. Furthermore, assessment of partial versus total hysterectomy was assessed only for two cohort studies. Elevated estimates are seen for both types, but significant findings are shown for partial hysterectomy only. The exact biological mechanism underlying the association between hysterectomy and kidney cancer is unclear, although both hormonal and iatrogenic factors are suggested [ 5 – 12 ]. Fluctuating sex hormones have been shown to make nephrons vulnerable to inflammation and oxidative stress [ 6 , 7 ]. Hormone replacement therapy, common among women who have undergone a hysterectomy, may adversely affect the kidneys as progesterone has been shown to inhibit the kidney’s ability to filter out toxins and estrogen-mediated cell proliferation in animals has been observed as an early event in the progression of estrogen carcinogenesis [ 31 , 32 ]. Anatomic changes after surgery has been suspected to lead to renal damage through increased lipid peroxidation, hydronephorsis, and renal obstruction [ 5 , 8 – 12 ]. Inadvertent injury to the ureter during surgery has been shown to lead to an increased incidence of post renal obstruction [ 33 ]. Persistent hydronephrosis without evidence of injury to the ureter, but possibly due to secondary pelvic anatomy changes post surgery, has also been reported in several prospective studies [ 34 , 35 ]. Furthermore, the effect of gynecological surgery has been shown to alter the rate of lipid peroxidation levels, which may induce DNA damage and promote mutations in proto-oncogenes and tumor suppressor genes [ 5 , 10 – 12 ]. The adverse association between hysterectomy and kidney cancer may be a consequence of the underling conditions leading up to a hysterectomy. In the US, the most common diagnosis associated with hysterectomy is uterine fibroids [ 36 ]. Studies have shown that genetic mutations involving fumarate hydratase, a Krebs cycle enzyme, predispose individuals to benign uterine fibroids as well as to hereditary kidney cancer [ 37 ]. The only study in our review to asses fibroids observed no association with kidney cancer risk [ 13 ]. This study was also the only to evaluate vaginal versus abdominal hysterectomies in relation to kidney cancer, where significantly elevated risk was observed only for abdominal hysterectomy. Abnormal uterine bleeding, endometriosis and prolapse are the second, third and fourth most commonly associated conditions associated with hysterectomies, respectively [ 36 ]. Although none of the studies investigated abnormal uterine bleeding, the relationship between endometriosis and kidney cancer was assessed in a series of epidemiological studies involving data from the National Swedish Inpatient Registry [ 38 – 40 ], the most recent which showed a significant 36% increase in risk [ 40 ]. For uterine prolapse, several case-reports have shown an association with acute and chronic renal failure, factors that increase the risk of kidney cancer [ 41 – 43 ]. Non-significant excess kidney cancer risk has been seen in patients with a previous ovarian cancer diagnosis [ 44 – 46 ]. No studies have evaluated the association between cesarean sections and kidney cancer risk. The carcinogenic potential of conditions that lead to a hysterectomy are poorly understood, but hypothesized to be multidimensional in etiology, involving hormonal, genetic, and immunological factors [ 39 ]. Limitations of our review must be acknowledged. Self-reported hysterectomy status was not verified with medical records for several studies included in our review. High agreement, however, has been reported between self-reported hysterectomy data (i.e. year of surgery) and hospital records in studies that validated exposure information [ 47 , 48 ]. For a number of cohort studies included in our review, exposure information was collected at baseline, and changes that may have occurred during follow-up were not considered. Yet, such misclassification was likely independent of future kidney cancer risk, and thus expected to bias association towards the null. The summary estimates from several subgroup meta-analyses are based on a small number of studies, and should be interpreted with caution. Also, while summary estimates were calculated separating out the effect of oophorectomy form hysterectomy, we were unable to tease out the possible influence of other gynecological procedures (e.g., history of cesarean sections, vaginal versus abdominal hysterectomy) and indications for hysterectomy (e.g., uterine fibroids, endometriosis, ovarian tumors, uterine prolapse), as this data was not provided in the vast majority of studies included in our review. Additional epidemiologic investigations with detailed data on such procedures and conditions could provide further insight into the relationship between hysterectomy and kidney cancer. Lastly, while our review was able to account for kidney cancer risk factors, results may have been confounded by factors related to poor lifestyle, such as low education level and lack of physical activity; although, a quick evaluation of studies in our review that had controlled for education level as well as established kidney cancer risk factors suggests that the hysterectomy association with kidney cancer is not affected by level of education. In conclusion, our findings from this meta-analysis suggest that women undergoing hysterectomy have an approximate 30% increased relative risk of kidney cancer. Although this relative risk is moderate in size, the proportion of kidney cancers among women attributable to hysterectomy may be substantial, given that approximately 45% of women are estimated to undergo this procedure by the age of 70 in the US [ 49 ]. Additional studies are needed to elucidate the biological mechanisms underlying this association.

Kidney cancer was the seventh- and ninth most commonly diagnosed malignancy among American men and women, respectively, in 2011 [ 1 , 2 ]. According to the International Agency for Research on Cancer, the incidence rate of this malignancy in 2008 for the United States (US) was one of the highest in the world, estimated at 16.1 per 100,000 men and 8.7 per 100,000 women [ 3 ]. Established risk factors for kidney cancer include: older age, male sex, family history of kidney cancer, cigarette smoking, excess body weight, and hypertension [ 4 ]. The difference in incidence by sex has motivated epidemiologic investigations into the etiologic relevance of hormonal and reproductive factors, the findings of which have been inconsistent. Recently, we assessed the role of reproductive factors on kidney cancer risk among post-menopausal women enrolled in two large US prospective cohort studies [ 5 ]. We found evidence suggesting that women who undergo hysterectomy have a 32% increased relative risk of future kidney cancer, possibly as a result of hormonal and/or iatrogenic effects [ 5 – 12 ]. Findings from previous epidemiologic studies have been inconsistent [ 13 – 26 ], with the majority of investigations having insufficient statistical power to detect an association of this magnitude. We conducted a systematic review and meta-analysis to summarize and better understand the published epidemiologic evidence regarding hysterectomy and kidney cancer risk.

We conducted a PubMed MEDLINE ( http://www.ncbi.nlm.nih.gov/sites/entrez ) search using the following key words: female reproductive factors and (kidney) cancer and hysterectomy and (kidney) cancer. All published cohort and case-control studies written in English from 1950 through 2012 that evaluated the association between hysterectomy and kidney cancer risk were included in our review. We also reviewed the references in all identified publications for additional studies. We calculated both overall and separately for cohort and case-control studies, summary relative risks (SRRs) and 95% confidence intervals (CIs) using random-effects models. For two case-control studies, risk estimates were not presented; where possible, we calculated crude risk estimates using the number of subjects provided in the manuscripts (i.e., unexposed/exposed subjects) [ 24 , 26 ]. Subgroup summary estimates were also calculated by study location (i.e., US or Europe; available for 11 studies), age at hysterectomy (<45 years, 45+ years; 5 studies), time since hysterectomy to enrollment (cohort) or diagnosis (case-control) (<10 years, 10+years; 5 studies), type of hysterectomy (i.e. total, partial, and restricted to those without an oophorectomy; 4 studies), and studies that computed risk estimates adjusted for adiposity, hypertension and smoking (6 studies). The Higgin’s I 2 statistic and Cochrane’s Q test were computed to evaluate heterogeneity across the studies [ 27 , 28 ]. Where heterogeneity was observed, each study was omitted one at a time from that analysis to identify the outlying study and to recalculate the summary risk estimate. We assessed publication bias statistically using the Egger’s and Begg’s methods as well as through evaluation of funnel plots [ 28 ]. The influence of potential publication bias on risk estimates was further evaluated by implementing the Duval and Tweedie nonparametric “trim-and-fill” method [ 29 ]. Statistical tests were two-sided with an alpha of 0.05. All analyses were conducted using STATA software Version 10 (College Station, TX).

Supplemental Figure 1. Funnel Plot of Standard Error by Log of the Risk Estimate for Studies of Hysterectomy Status and Kidney Cancer Risk.