The Association of Kidney Function and Inflammatory Biomarkers with Epithelial Ovarian Cancer Risk.

In total, 232,908 women were included in the analyses. Most (68%) had clinically normal eGFR crcys levels (≥90 mL/min/1.73m 2 ). Women with clinically normal eGFR crcys were younger and slightly more likely to be premenopausal compared to women with lower levels of eGFR crcys ( Table 1 ). Further, women with eGFR crcys ≥90 were diagnosed at a younger age, were less likely to have a tubal ligation or hysterectomy, more likely to use OCs, and less likely to be nulliparous, but more likely to have two or three children than women with lower levels of eGFR crcys . Women with lower versus higher eGFR crcys levels were more likely to be current smokers and use blood pressure medication. Overall, there were no statistically significant associations between any kidney marker measured in blood ( Table 2 ) or urine ( Table 3 ) and ovarian cancer risk after Bonferroni correction. Further adjustment for covariates and CRP did not substantially alter the results. For example, comparing the top versus bottom quartile of eGFR crcys , the HR was 1.00 (95%CI=0.83, 1.22). There was a nominally significant trend for increased risk of ovarian cancer with increased CRP quartiles (Q4 vs. Q1: HR: 1.16, 95%CI: 0.96, 1.41; p-trend=0.03). Using cutpoints from prior studies, CRP 3-<10 mg/L versus 10 vs. 3 mg/L versus ≤3 mg/L was associated with an increased risk of ovarian cancer (HR: 1.21, 95%CI: 1.05, 1.40). There were no significant associations between eGFR crcys , creatinine, cystatin C, uric acid, ( Supplemental Table 2 ) or microalbumin ( Supplemental Table 3 ) and ovarian cancer risk by histologic subtype. Uric acid (measured in serum) was nominally positively associated with risk of endometrioid ovarian cancer (Q2 vs. Q1: HR: 2.45, 95%CI: 1.07, 5.63; Q3 vs. Q1: HR: 2.24, 95%CI: 0.95, 5.26; Q4 vs. Q1: HR: 1.36, 95%CI: 0.52, 3.54) and inversely associated with risk of clear cell ovarian cancer (Q2 vs. Q1: HR: 0.77, 95%CI: 0.38, 1.57; Q3 vs. Q1: HR: 0.31, 95%CI: 0.12, 0.81; Q4 vs. Q1: HR: 0.48, 95%CI: 0.20, 1.12). Additionally, there was a trend of lower clear cell ovarian cancer risk with higher serum uric acid levels (p=0.04), but no statistically significant heterogeneity by histologic subtype (p=0.21). Women with potassium levels (measured in urine) in the highest versus lowest quartile had a lower risk of endometrioid ovarian cancer (HR: 0.33, 95%CI: 0.11, 0.98, p-trend: 0.02) and an increased risk of clear cell ovarian cancer (HR: 4.74, 95%CI: 1.39, 16.16, p-trend: 0.01). This finding was suggestively different by histologic subtype (p=0.005). There were no statistically significant associations between markers in the blood by CRP levels ( Supplemental Table 4 ); however, urinary sodium was nominally associated with increased ovarian cancer risk for women with CRP >3.0 mg/L (p-trend=0.03), but not for women with CRP ≤3 mg/L (p-interaction=0.01; Supplemental Table 5 ). Although not statistically significant, some markers had a suggestive increased risk of ovarian cancer for individuals with higher CRP (e.g., uric acid Q4 vs. Q1 HR, high CRP: 1.23, 95%CI: 0.81, 1.86; HR, low CRP: 0.83, 95%CI: 0.66, 1.05).

We included women in the UKB, a longitudinal prospective cohort study that recruited over 500,000 men and women (n=270,737) aged 40–69 years between 2006 and 2010. Participants lived in England, Wales, and Scotland and attended an assessment center visit to fill out a touchscreen questionnaire, get physical measurements, and collect biological samples. Participant data was linked to national death registries, cancer registries, and inpatient hospital admissions. This study was approved by the North West Multi-centre Research Ethics Committee in the UK, and the institutional review boards (IRB) of Moffitt Cancer Center (Advarra IRB Pro00022823) and Nova Southeastern University (IRB 2022–509). Ovarian cancer was determined using ICD-10 codes from the Health and Social Care Information Centre for UKB participants living in England and Wales and through the National Health Service for participants living in Scotland. Cases included ICD-10 codes for ovary (C561, C562, C569), fallopian (C570) and peritoneal (C451, C481, C482, C488) cancers. We considered individual associations for serous, endometrioid, clear cell, and mucinous ovarian histological subtypes, collected from the UK Cancer Registry. Biomarkers of kidney function were measured from serum and spot urine samples collected at study baseline (described previously in ( 19 )) and were assayed in a central laboratory from 2014–2017. Measurement of creatinine (serum and urine), urea (serum), and uric acid (serum) was conducted using enzymatic analysis (serum: Beckman Coulter AU5800, urine: Beckman Coulter AU5400). CRP and cystatin C were measured by immunoturbidimetric analysis (serum; Beckman Coulter AU5800 [CRP]; Siemens Advia 1800 [cystatin C]). Microalbumin was measured by immunoturbidimetric analysis and potassium and sodium were measured by ion selective electrode analysis (urine; Beckman Coulter AU5400). Coefficients of variation for all biomarkers were ≤3% across all assay batches ( 20 ). Estimated glomerular filtration rate (eGFR crcys ) was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine-cystatin equation ( 21 – 24 ). A previous study examined the long-term reproducibility of these biomarkers within individuals over time ( 25 ). The intraclass correlation coefficients (ICCs) for biomarkers of kidney function in serum ranged from 0.58 for urea to 0.82 for uric acid, and CRP was 0.29. Biomarkers in urine had lower ICCs ranging from 0.26 for potassium to 0.41 for microalbumin. Body mass index (BMI) was calculated from height and weight measured at the initial assessment visit. Blood pressure medication use (yes, no), family history of breast cancer (mother and/or sister: yes, no), fasting status (yes, no), hormone therapy use (ever, never), hysterectomy (yes, no), menopausal status (premenopausal, postmenopausal, unknown), oral contraceptive (OC) use (never, <1 -<5, 5-<10, ≥10 years, unknown), parity (none, one, two, three, four or more children), smoking status (never, past smoker, current smoker), and tubal ligation (yes, no) were collected from self-reported questionnaire or interview data at baseline. Time of sample collection was recorded (12:00am-11:00am, 11:01am-2:00pm, 2:01pm-4:00pm, 4:01pm-11:59pm). Women without a report of cancer diagnosis (except non-melanoma skin cancer) or bilateral oophorectomy prior to baseline and were not missing date of birth or date of death (if deceased) were included (n=234,844). Women who were missing all biospecimen data (n=1,887) or who had a non-epithelial ovarian cancer diagnosis (n=49) were excluded. We created quartile cut points based on the entire eligible UKB cohort and estimated hazard ratios (HR) and 95% confidence intervals (CI) of quartiles of each biomarker with ovarian cancer risk using Cox proportional hazard models. Tests for trend were calculated using the median of each marker quartile. Models were initially stratified by age and assessment center, then additionally adjusted for the covariates as listed above, and finally additionally including CRP (quartiles). We conducted competing risk Cox proportional hazard models to assess HRs and 95% CIs by ovarian histological subtype. To evaluate effect modification by CRP (≤3.0 vs. >3.0 mg/L), we conducted stratified Cox proportional hazard models using multiplicative interaction terms and likelihood ratio tests to calculate p-values for interaction. As the association of CRP with ovarian cancer risk has not been reported for the UKB, we conducted Cox proportional hazard models of CRP (<1, 1-<3, 3-<10, ≥10 mg/L, based on prior studies ( 5 , 26 , 27 )) as described above. We used a nominal alpha-level of 0.05; however, due to multiple testing and the correlated nature of the kidney markers, we calculated the effective number of independent tests to be 7 ( 28 ), and considered a Bonferroni corrected p-value of 0.007 (0.05/7) for these markers. Given a priori literature supporting an association of CRP with ovarian cancer risk, this was evaluated at the alpha-level of 0.05. All analyses were conducted in SAS (v9.4, Cary, North Carolina). The data used in this study are available by open application through the UK Biobank.

This is the first study to evaluate biomarkers of kidney function and ovarian cancer risk as well as assess potential effect modification by inflammation. In general, there were no clear associations between any marker of kidney function and ovarian cancer risk overall or by histology. There was a suggestion that associations for uric acid and potassium varied by histologic subtype; however, power was limited. Consistent with prior studies, we observed a higher risk of ovarian cancer with increasing CRP levels ( 5 , 26 , 27 ). Finally, we observed suggestively stronger associations of some kidney markers among women with higher versus lower CRP, although most associations did not reach nominal statistical significance. Contrary to our hypothesis, we did not observe associations between kidney function markers and risk of ovarian cancer. This may be because many individuals were within the normal range, reducing power to detect associations for those with more extreme phenotypes. Prior work has shown that participants of the UKB are generally healthier than the UK population ( 29 ). For example, in the UK general population, women have about a 7 to 9% prevalence of stage three or greater chronic kidney disease (CKD; eGFR crcys <60 mL/min/1.73m 2 ) compared to 1.4% in our study population ( 30 , 31 ). Another potential reason is the differential effects of estrogen on ovarian cancer risk and kidney disease. Endogenous and exogenous estrogen use has been shown to increase ovarian cancer risk for certain histologies, but higher estrogen is related to reduced risk and slower progression of CKD ( 32 – 34 ). Although this study controlled for postmenopausal hormone use, we were unable to adjust for type of exogenous hormone used (e.g., estrogen, progesterone) or endogenous levels of estrogen. Further, women with CKD experience menopause on average 4.5 years earlier than women without CKD, suggesting they may have a reduced number of lifetime ovulatory cycles ( 35 ). Fewer lifetime ovulatory cycles are related to lower ovarian cancer risk, but higher CRP levels; thus, our inability to control precisely for this factor may have led to residual confounding ( 36 , 37 ). Future work should evaluate if kidney function may differentially affect risk of ovarian cancer by estrogen levels. There were no significant interactions between CRP and markers of kidney function with ovarian cancer risk, although there were suggestive trends that worse kidney function may be related to higher risk among those with elevated CRP levels. Measures of inflammation (including CRP) are negatively correlated with kidney function and can increase risk of rapid kidney decline ( 13 , 38 ). Further, prior work has observed increased risk of ovarian cancer with CRP levels >10 mg/L vs. <1 mg/L ( 5 ), while we only observed increased risk of ovarian cancer with CRP levels 3-<10 mg/L vs. <1 mg/L and ≥3 mg/L vs. 10 mg/L vs. <1 mg/L) was associated with a significant increase in ovarian cancer risk only in women who used OCs ( 5 ). Interestingly, OCs can worsen kidney function through increased activation of the renin-angiotensin-aldosterone system ( 39 – 41 ). Further, OC use has been associated with increased CRP levels but a decreased risk of ovarian cancer ( 1 , 42 – 44 ). This suggests there is a complex interconnection between kidney function, ovarian cancer risk, inflammation, and OCs that should be investigated further. This study had several strengths including a very large cohort (over 230,000 women) and a long follow up period (approximately one million person-years). The UKB has a breadth of information on reproductive and health factors (e.g., OC use, parity, etc.), which permitted control for potential confounders. Finally, this study used data from clinical assays that have demonstrated reliability and validity ( 20 , 45 ), improving interpretability of the results. Although the assays used have been previously validated, the ICCs of markers in urine were 0.41 or less ( 25 ), suggesting results with these markers should be interpreted cautiously as women had measures at one time that may not accurately reflect long term levels. The UKB is a mostly non-Hispanic white cohort that tends to be healthier than the general UK population ( 29 ) which limits the generalizability of the study to other racial and ethnic groups and to those with poorer health. Several variables were not available in the UKB dataset such as family history of ovarian cancer and grade of the tumor . This limited the ability to control for a potential confounder and evaluate differences by high and low-grade tumors. Overall, this study provided the first evaluation between markers of kidney function and ovarian cancer risk. No clear associations were observed overall, by histological subtype, or by CRP level, although there were suggestive trends for some markers indicating that worse kidney function may impact risk of specific histologic subtypes or increase risk of ovarian cancer only among women with high CRP levels. This study was limited in the ability to detect modest associations because fewer than expected women had marker values outside the normal range; therefore, future work is needed to evaluate this association in cohorts that include more individuals with chronic health conditions. Additionally, exogenous and endogenous estrogens have a complex relationship with inflammation, kidney function, and ovarian cancer development, suggesting that future work should evaluate the potential inter-relationships of these factors.

Ovarian cancer is the seventh most diagnosed female cancer worldwide, and few modifiable risk factors have been identified, limiting opportunities for prevention ( 1 , 2 ). A well supported mechanism of ovarian tumorigenesis is through inflammatory pathways ( 3 , 4 ). C reactive protein (CRP) is a nonspecific marker of systemic inflammation. CRP levels >10 mg/L have been linked to a 67% increased risk for ovarian cancer (95% CI: 1.12–2.48) compared to woman with CRP <1 mg/L ( 5 ). Other inflammation-related risk factors for ovarian cancer include obesity, endometriosis, and chlamydia, which can cause increased local or systemic inflammation ( 6 – 8 ). Notably, CRP has been shown to increase risk of chronic kidney disease in several populations ( 9 – 12 ). Moreover, individual markers of kidney function such as albumin, creatinine, uric acid, cystatin C and estimated glomerular filtration rate (eGFR) have been independently associated with increased inflammatory markers such as CRP, tumor necrosis factor α (TNF-α), and interleukin-6 (IL-6) ( 13 – 17 ). In addition, a prior study reported a modest increased risk in ovarian cancer among current users of thiazide diuretics, which can reduce kidney function, compared to never users (Hazard Ratio [HR] 1.37, 95% Confidence Interval [CI]: 1.13, 1.69) ( 18 ), further suggesting impaired kidney function as a potential risk factor for ovarian cancer. To our knowledge, no study has evaluated the association between markers of kidney function and ovarian cancer risk. Therefore, we conducted a study considering multiple serum and urinary biomarkers of kidney health, including creatinine, cystatin C, urea, uric acid, microalbumin, potassium, and sodium, as well as CRP in relation to ovarian cancer risk overall and by histological subtype in the UK Biobank (UKB). Additionally, given the intersection of kidney function and inflammation, we evaluated potential effect modification of the associations between kidney function markers and risk of ovarian cancer by CRP.