Additive interaction of diabetes mellitus and chronic renal failure in the mortality risk of cancer patients

Study design and population

The study population consisted of cancer patients over the age of 20 who visited Samsung Medical Center (SMC), Seoul, Republic of Korea from January 2008 to December 2019. A total of 138,956 subjects who were diagnosed with cancer for the first time (International Classification of Disease, 10th revision; (ICD-10), code C) and had records with TNM stage were enrolled. We stratified these cancer patients according to the presence or absence of DM and CKD. We excluded subjects without a medical history of DM based on self-report or without laboratory test results of hemoglobin A1c (HbA1c) or fasting blood glucose (n = 15,150). Subjects with a history of type 1 diabetes with an ICD-10 code of E10 (n=30), subjects with missing values ​​for serum creatinine (n=2520), who had kidney cancer (n=3442), who had distant metastases (n = 9031), and who had missing variables of BMI, history of alcohol consumption, smoking, and hypertension (n = 7099) were excluded. Finally, a total of 101,684 subjects were analyzed (Supplementary Fig. 1).

For further analysis to assess the effect of DKD on mortality risk in cancer patients with DM, subjects who had laboratory results of urine albumin to creatinine ratio (UACR) were included among the cancer patients with a pre-existing DM cluster (n=2384).

Definition of exposure

Data were extracted from SMC’s Clinical Data Warehouse (CDW) DARWIN-C for this study. Personal medical history, including diabetes, hypertension, smoking history, alcohol consumption, and medications, was assessed by electrical medical records and a self-administered questionnaire. All demographic, anthropometric and laboratory data of the subjects were collected. DM was defined as having ICD-10 codes E11-14, self-reported history of DM, antidiabetic medication prescription records, HbA1c of 6.5% or greater, or fasting blood glucose of 126 mg/ dL or more. The estimated glomerular filtration rate (eGFR) value was calculated using the formula from the Epidemiology of Chronic Kidney Disease Collaboration 2021 (CKD-EPI 2021)31. The definition of the presence of IRC was eGFR 2 at the diagnosis of cancer32. DKD was defined as an eGFR 2 or UACR ≥ 30 mg/g at diagnosis of cancer in patients with diabetes mellitus13.

Definition of covariates

Body weight and height were measured and body mass index (BMI) was calculated as body weight (kg) divided by height squared (m2). Smoking status and alcohol consumption were collected through a self-reported questionnaire and categorized as never, never, or current. History of hypertension was defined as having a self-reported history of hypertension, having I10-15 in ICD-10 codes, having records of antihypertensive drug prescriptions, or having at least 3 times systolic blood pressure greater than 140 mmHg or 90 mmHg in diastolic blood pressure. According to the original cancer site, we have divided all cancers into 24 common categories33and reclassified them into 8 types of cancer, including gastrointestinal (colon, rectum, stomach, esophagus and small intestine), urological (bladder, prostate, testicle and ureter), gynecological (endometrium, cervix, body uterine and ovarian), breast cancer, hepato-pancreatobiliary (liver and intrahepatic bile ducts, gallbladder and other parts of the bile ducts and pancreas), lung cancer, thyroid cancer and other cancers11.12.

Definition of results

All patients were followed from the date of their first cancer diagnosis until the end of the study (December 2019) or death (collected from SMC CDW death records linked to Statistics Korea).

statistical analyzes

All continuous variables were presented as mean and standard deviation (SD) and all categorical variables were presented as proportions. Analysis of variance (ANOVA) for continuous variables and chi-square test for categorical values ​​were used to assess characteristics based on the presence of DM or IRC. The survival curves were analyzed by the Kaplan-Meier method and compared to the log-rank test. We assessed hazard ratios (HR) with 95% confidence interval (CI) for all-cause mortality using Cox proportional hazard regression models. To perform a multivariate analysis, we adjusted for age, sex (male and female), BMI (218.5–22.9 kg/m223–24.9kg/m2≥ 25kg/m2), alcohol consumption (never, never, current), smoking (never, never, current), history of hypertension (yes, no) and type of cancer (gastrointestinal, urological, gynecological, breast, hepato-pancreatobiliary , pulmonary, thyroid cancer and others). Next, we performed additive interaction analyzes of DM and CRI on mortality risk by estimating additive interaction parameters, including excess relative risk due to interaction (RERI), attributable proportion due to interaction (AP) and synergy index (SI)34,35,36,37. We identified the additive interaction stratified by gender (male, female), age (2≥ 25kg/m2). We confirmed the three-way interaction effect of CKD and DM with each subgroup and the two-way interaction effect of DKD and gender in cancer patients with preexisting DM. Finally, we performed a sensitivity analysis to test the robustness of our study. Considering cancer patients who achieved No Evidence of Disease (NED) status, which was determined by oncologists during follow-up, they were censored at the start date of NED instead of continuing follow-up until at the end of the study38.Statistical significance was considered as P

Ethical approval and consent to participate

The Institutional Review Board (IRB) of Samsung Medical Center has approved this study (SMC Approval No. 2021-08-092). An informed consent waiver was granted by the IRB because all data provided by the MSC CDW to researchers was anonymized and released for research purposes. All methods were conducted in accordance with the Declarations of Helsinki.