Metformin Treatment Among men With Diabetes and the Risk of Prostate Cancer

A Population-Based Historical Cohort Study

Laurence S. Freedman; Nirit Agay; Ruth Farmer; Havi Murad; Liraz Olmer; Rachel Dankner

Am J Epidemiol. 2022;191(4):626-635.

Abstract and Introduction

Abstract

There is conflicting evidence regarding the association between metformin treatment and prostate cancer risk in diabetic men. We investigated this association in a population-based Israeli cohort of 145,617 men aged 21–89 years with incident diabetes who were followed over the period 2002–2012. We implemented a time-dependent covariate Cox model, using weighted cumulative exposure to relate metformin history to prostate cancer risk, adjusting for use of other glucose-lowering medications, age, ethnicity, and socioeconomic status. To adjust for time-varying glucose control variables, we used inverse probability weighting of a marginal structural model. With 666,553 person-years of follow-up, 1,592 men were diagnosed with prostate cancer. Metformin exposure in the previous year was positively associated with prostate cancer risk (per defined daily dose; without adjustment for glucose control, hazard ratio (HR) = 1.53 (95% confidence interval (CI): 1.19, 1.96); with adjustment, HR = 1.42 (95% CI: 1.04, 1.94)). However, exposure during the previous 2–7 years was negatively associated with risk (without adjustment for glucose control, HR = 0.58 (95% CI: 0.37, 0.93); with adjustment, HR = 0.60 (95% CI: 0.33, 1.09)). These positive and negative associations with previous-year and earlier metformin exposure, respectively, need to be confirmed and better understood.

Introduction

There is conflicting evidence regarding the effect of metformin therapy on prostate cancer risk. Recent meta-analyses^[1–3] of observational studies found no clear evidence of a previously hypothesized protective association.^[4–7] Nevertheless, analyzing observational study data addressing this question is fraught with pitfalls^[8] that, in turn, can influence meta-analysis results, so the question remains open. The high prevalence of diabetes, widespread use of metformin treatment for diabetes, and relatively high incidence of prostate cancer make this question important.

We describe here analysis of a population-based cohort study of patients diagnosed with diabetes, aimed at addressing this question. Important features of our analysis are the use of Cox regression with time-dependent covariates describing metformin treatment history^[9] and inverse probability weighting (IPW) of marginal structural models (MSMs).^[10] MSM analysis addresses bias arising in Cox regression when a time-varying treatment is modified in response to a time-varying marker—here, hemoglobin A1c (HbA1c) or blood glucose level—that is itself associated with the disease outcome, prostate cancer.

1 2 3 4

Next Section

Abstract and Introduction
Methods
Results
Discussion
References

Table 1. Baseline Characteristics of All Men Insured by an Israeli Health Maintenance Organization With Incident Diabetes During 2002–2012 and Those Included in the Final Analysis
Characteristic	All Men With Incident Diabetes (n = 193,835)	Men With Incident Diabetes Included in the Study^a (n = 145,617)
Age at diagnosis, years^b	60.9 (14.1)	60.9 (13.1)
Ethnic origin, %
Ashkenazi Jew	31.7	30.1
Sephardic Jew	27.9	28.2
Israeli-born Jew	18.2	18.9
Israeli Arab	16.7	17.3
Yemenite, Ethiopian, or Central African	5.4	5.5
Socioeconomic status^c, %
Low	42.3	42.6
Medium	37.9	38.0
High	17.0	16.6
Missing data	2.8	2.8
Cigarette smoking, %
Never smoker or missing data	49.5	47.3
Past smoker or current smoker	50.5	52.7

^aMen with incident diabetes who had at least 2 years of follow-up before prostate cancer incidence, death, or reaching age 90 years.
^bValues are expressed as mean (standard deviation).
^cDetermined by locality of the Clalit Health Services clinic (Tel Aviv, Israel).

Table 2. Course of Glucose-Lowering Medication Use Among Israeli Men With Incident Diabetes and At Least 2 Years of Follow-up ( n = 145,617), 2002–2012 ^a
Type of GLM	First Therapy				Second Therapy
	No. of Men	% of Subtotal	Time From Diagnosis to Start of First Therapy, no. of quarters		No. of Men	% of Subtotal	Time From First Therapy to Second Therapy, no. of quarters
	No. of Men	% of Subtotal	Mean	Median (IQR)	No. of Men	% of Subtotal	Mean	Median (IQR)
Metformin	71,493	75.21	6.7	3 (0–11)	7,147	19.38	9.8	7 (3–15)
Sulfonylurea	8,052	8.47	4.2	1 (0–6)	16,242	44.03	10.5	9 (4–15)
Repaglinide	2,314	2.43	7.1	3 (0–11)	6,436	17.45	11.5	10 (4–17)
Insulin	1,337	1.41	6.3	2 (0–10)	3,590	9.73	13.7	12 (6–20)
Acarbose	629	0.66	4.9	2 (0–7)	960	2.60	9.7	8 (3–14)
GLP1	27	0.03	10.0	10 (0–19)	402	1.10	15.8	14 (8–23)
DPP4i	134	0.14	15.3	13 (5–23)	1,455	3.94	14.6	14 (6–21)
Rosiglitazone	70	0.07	5.4	2.5 (0–8)	655	1.78	10.9	10 (5–16)
Metformin + sulfonylurea	6,898	7.26	4.0	0 (0–5)
Insulin + sulfonylurea	114	0.12	3.8	0 (0–6)
Insulin + metformin	679	0.71	5.2	1 (0–8)
Other combination	2,377	1.63	6.0	2 (0–9)
≥3 GLMs	935	0.64	6.3	1 (0–10)
Subtotal	95,059		6.3	3 (0–10)	36,887		11.1	9 (4–16)
None	50,558		25.1^b	24 (16–34)^b	53,205		16.7^b	15 (8–24)^b
Total	145,617				95,059^c

Abbreviations: DPP4i, dipeptidyl peptidase-4 inhibitor; GLM, glucose-lowering medication; GLP1, glucagon-like peptide-1 receptor agonist; IQR, interquartile range.
^aMedication data were obtained for the period 1998–2011 (60 quarter-years). The study started on January 1, 2002, and follow-up continued until prostate cancer diagnosis, death, age 90 years, or December 31, 2012, whichever occurred first.
^bTime for "none" represents time to the end of follow-up (in quarter-years).
^cIn addition, 4,967 men had intensification of treatment with a combination of medications or the addition of ≥1 other medications from the same group (e.g., 2 types of insulin).

Table 3. Hazard Ratios for the Association of Prostate Cancer With Metformin Exposure (per 1 Defined Daily Dose per Day) in Various Time Periods During the 7 Years Prior to Cancer Diagnosis Among Israeli Men (Derived From a Cox Model), 2002–2012
Period of Metformin Exposure^a	HR^b	95% CI
Previous year	1.53	1.19, 1.96
Second–fourth years before	0.62	0.41, 0.94
Fifth–seventh years before	0.94	0.55, 1.60
Second–seventh years before^c	0.58	0.37, 0.93

Abbreviations: CI, confidence interval; HR, hazard ratio.
^aPer 1 defined daily dose of metformin per day over the specified period.
^bAdjusted for age (in 5-year subgroups), race/ethnicity, socioeconomic status, and history of use of other glucose-lowering medications (in 4 groups: insulins; insulin secretagogues (sulfonylureas, meglitinides) and incretin mimetics (dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists); α-glucosidase inhibitors; and rosiglitazone (i.e., thiazolidinediones)).
^cDerived from the HRs for the second–fourth years before and the fifth–seventh years before, as follows: HR_2–7 = HR_2–4 × HR_5–7.

Table 4. Odds Ratios for the Association of Prostate Cancer With Metformin Exposure (per 1 Defined Daily Dose per Day) in Various Time Periods During the 7 Years Prior to Cancer Diagnosis Among Israeli Men (Derived From Marginal Structural Models), 2002–2012
Period of Metformin Exposure^a	Unweighted MSM^b		Weighted MSM^c
Period of Metformin Exposure^a	OR	95% CI	OR	95% CI
Missing-Value Indicators Method (n = 145,617)
Previous year	1.57	1.20, 2.06	1.42	1.04, 1.94
Second–fourth years before	0.65	0.41, 1.01	0.73	0.44, 1.20
Fifth–seventh years before	0.70	0.38, 1.28	0.83	0.44, 1.56
Second–seventh years before	0.45	0.26, 0.77	0.60	0.33, 1.09
Last-Value-Carried-Forward Method (n = 105,412)
Previous year	1.41	1.05, 1.89	1.41	1.05, 1.91
Second–fourth years before	0.72	0.45, 1.16	0.77	0.48, 1.23
Fifth–seventh years before	0.62	0.33, 1.19	0.63	0.33, 1.19
Second–seventh years before	0.45	0.25, 0.80	0.48	0.27, 0.87
Time-Sequential Imputation Method (n = 145,614)
Previous year	1.57	1.20, 2.06	1.62	1.19, 2.20
Second–fourth years before	0.65	0.41, 1.01	0.67	0.42, 1.08
Fifth–seventh years before	0.70	0.38, 1.28	0.72	0.39, 1.34
Second–seventh years before	0.45	0.27, 0.77	0.49	0.27, 0.86

Abbreviations: CI, confidence interval; OR, odds ratio; MSM, marginal structural model.
^aPer 1 defined daily dose of metformin per day over the specified period.
^bAdjusted for baseline age (in 5-year subgroups), race/ethnicity, socioeconomic status, and time since diabetes diagnosis.
^cAlso adjusted for blood glucose level and hemoglobin A1c level (time-varying confounders).