#840 Kidney cancer, treatment with nephrectomy and survival in chronic kidney disease (CKD)

Abstract Background and Aims People with chronic kidney disease (CKD) have increased incidence of kidney cancer and poorer survival post-diagnosis compared to people without CKD. Nephrectomy, either partial or radical is the primary treatment for many kidney cancers and can be curative even in cases with locally advanced disease or single metastasis. We hypothesised that nephrectomy may be less frequently undertaken in people with kidney cancer and pre-existing CKD and that this might contribute to poorer survival in this group. Method Using data from the prospective research cohort UK Biobank (n = 502,536) linked to cancer registry, participants with available biochemistry and a new diagnosis of kidney cancer during follow-up were identified. Occurrence of nephrectomy was extracted from linked hospital data and all-cause mortality data from death certification records. Estimated glomerular filtration rate (eGFRCr) was calculated by CKD-EPI 2009 formula with race coefficient, as this was the formula recommended for use in the UK at the time of recruitment to UK Biobank. Cox proportional hazards models tested associations between eGFRCr and i) undergoing nephrectomy and ii) all-cause mortality by 10 ml/min/1.73 m2 decline from a reference value of 90 ml/min/1.73 m2. We additionally conducted subgroup analyses for nephrectomy and all-cause mortality within eGFRCr categories 105, 90-105, 75-90 (reference), 60-75, 45-60, 30-45 and 30 ml/min/1.73 m2. Both models were adjusted for age, sex, diabetes mellitus, hypertension, stroke, ischaemic heart disease, body mass index and socioeconomic deprivation. The all-cause mortality model was additionally adjusted for undergoing nephrectomy. Results There were 2053 participants with a new diagnosis of kidney cancer and sufficient biochemical data for inclusion. Of these, 935 (46%) had eGFRCr 60-90 ml/min/1.73 m2 and 139 (7%) had eGFRCr 60 ml/min/1.73 m2. Median time from baseline to kidney cancer diagnosis was 5.8 (IQR 0.9, 8.8) years. Median follow-up time after a diagnosis of kidney cancer was 3.4 (IQR 0.9, 6.6) years. For each 10 ml/min/1.73 m2 decline in eGFRCr below 90 ml/min/1.73 m2, there was no significant difference in likelihood of receiving nephrectomy (HR 1.05, p = 0.08) and no significant difference in likelihood of all-cause mortality (HR 1.03, p = 0.23). People with moderate-advanced CKD had increased hazards of all-cause mortality compared to the reference group (eGFRCr 75-90 ml/min/1.73 m2) when models were run by eGFRCr category (eGFRCr 30-45—HR 1.9 95% CI 1.3-3.0; eGFRCr 30—HR 2.0 95% CI 1.0-4.0). Conclusion In UK Biobank participants with kidney cancer and mild CKD, there was no significant difference in likelihood of nephrectomy or in hazards of all-cause mortality associated with declining eGFRCr as a continuous variable. Moderate to advanced CKD (eGFR 45) was associated with increased all-cause mortality despite adjustment for nephrectomy, suggesting that this finding is not explained by reduced access to surgical treatment.