ABSTRACT
Aims/Introduction
A recent US Food and Drug Administration report highlighted concerns over nitrosamine (7‐nitroso‐3‐(trifluoromethyl)‐5,6,7,8‐tetrahydro[1,2,4] triazolo‐[4,3‐a]pyrazine [NTTP]) impurities in sitagliptin, prompting investigations into its safety profile. The present study aimed to determine if the use of NTTP‐contaminated sitagliptin, in comparison with other dipeptidyl peptidase‐4 (DPP‐4) inhibitors, is associated with an increased cancer risk.
Materials and Methods
This retrospective cohort study secondarily used the National Database of Health Insurance Claims and Specific Health Checkups of Japan, encompassing data on >120 million individuals. The study involved patients who initiated DPP‐4 inhibitor therapy (sitagliptin or other DPP‐4 inhibitors) and continued its exclusive use for 3 years. Sitagliptin users were compared with other DPP‐4 inhibitor users for assessing the occurrence of cancers, as defined by diagnosis codes. Further analyses focused on specific types of cancer, using either diagnosis codes or a combination of diagnosis and procedure codes. We also carried out various sensitivity analyses, including those with different exposure periods.
Results
Sitagliptin users (149,120 patients, 388,356 person‐years) experienced 9,643 cancer incidences (2,483.0/100,000 person‐years) versus 12,621 incidences (2,504.4/100,000 person‐years) among other DPP‐4 inhibitor users (199,860 patients, 503,952 person‐years), yielding a minimal difference (incidence rate ratio 0.99, 95% confidence interval 0.97–1.02). A multiple Cox proportional hazards model showed no significant association between sitagliptin use and overall cancer incidence (hazard ratio 1.01, 95% confidence interval 0.98–1.04). Findings were also consistent across cancer types and sensitivity analyses.
Conclusions
We observed no evidence to suggest an increased cancer risk among patients prescribed NTTP‐contaminated sitagliptin, although continued investigation is needed.
Keywords: Nitrosamines, Pharmacoepidemiology, Sitagliptin phosphate
This study investigated the incidence of cancer among 7‐nitroso‐3‐(trifluoromethyl)‐5,6,7,8‐tetrahydro[1,2,4] triazolo‐[4,3‐a]pyrazine‐contaminated sitagliptin users compared with other dipeptidyl peptidase‐4 inhibitors. As a result, the incidence was almost the same.
INTRODUCTION
Approximately 5 years ago, dipeptidyl peptidase 4 (DPP‐4) inhibitors were the fourth most commonly used antidiabetic medication in the USA 1 , and most commonly used antidiabetic medication in Japan 2 . In August 2022, the US Food and Drug Administration (FDA) revealed the contamination of sitagliptin, the first‐in‐class DPP‐4 inhibitor, with 7‐nitroso‐3‐(trifluoromethyl)‐5,6,7,8‐tetrahydro[1,2,4] triazolo‐[4,3‐a]pyrazine (NTTP), but the origin was not well‐known. NTTP is a form of nitrosamine, which is identified as a possible carcinogen 3 . Simultaneously, the FDA raised the threshold for an acceptable intake limit of NTTP from 37 to 246.7 ng/day, due to concerns regarding a potential sitagliptin shortage. The FDA evaluated that the additional risk associated with the elevated threshold was minimal 3 . This evaluation was based on the FDA's standard approach to determining acceptable intake limits, which aims to maintain a one in 100,000 cancer risk over 70 years of exposure 4 .
However, currently, no reports exist on whether using sitagliptin, compared with other similar medications, is associated with an increased cancer risk. Therefore, the present study aimed to investigate the cancer incidence associated with the use of sitagliptin in comparison with other medications using the national electronic claims database in Japan.
MATERIALS AND METHODS
Design, setting and participants
The present retrospective cohort study utilized the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB; Data S1) 5 . NDB data between April 2012 and March 2021, which were the medical contact dates, were extracted and used for this study.
Figure S1 provides an overview of the longitudinal study design used in the present study. Among patients who had a regular prescription (at least 1 prescription within 3 consecutive months) for the targeted medication, we excluded patients with a time 0 of March 2021, those whose cancer care started on or before time 0 and those who were aged <20 years at time 0.
Our study proposal was approved by the Institutional Review Board of the National Center for Global Health and Medicine (NCGM‐G‐004081‐06). We adhered to the NDB data user guidelines. Informed consent from participants was not obtained, as the data were anonymized before our receipt.
The definition of each variable, including exposure variables, outcome variables and covariates, as well as those stated in the inclusion and exclusion criteria mentioned above, is presented in Table S1.
Exposure variables
The main exposure variable in the study was the initial and exclusive use of a specific type of antidiabetic medication for a defined period (e.g., 3 years in the main analysis). The exposure variable of interest was the use of sitagliptin, which was supposedly contaminated with NTTP. All sitagliptin‐containing products on the market until 2021 were considered to be contaminated. In contrast, the control in the main analysis was the use of DPP‐4 inhibitors other than sitagliptin. Modifications in dosage and exchanges within the DPP‐4 inhibitors, excluding sitagliptin, were allowed. Furthermore, as described above, the additional analysis incorporated two additional medications, sulfonylurea and metformin, as controls.
Outcome variables
In the main analysis, the outcome variable was cancer incidence, identified through diagnosis codes in the claims data. The validity of the diagnoses of claims data for identifying cancer incidence was good (93.6% sensitivity and 97.5% specificity) in a study from the Japan Public Health Center‐based Prospective Study for the Next Generation 6 .
Five major cancer types in Japan were defined: stomach, colorectal, liver, lung and breast cancer (limited to women) from the diagnosis codes in the claims. Information on the combination of these five major cancer types with relevant procedures and medications was also collected to enhance the specificity of cancer incidence. In addition, each cancer type other than the five major cancers was defined based on diagnosis codes. The timing of cancer incidence was determined by the start date of the cancer diagnosis code in the claims data.
Covariates
To stratify and control for covariates, information was extracted from the claims. This included sex, age (categorized into groups, such as 20–39 years, 40–44 years, followed by subsequent 5‐year intervals, extending to ≥85 years), prefecture, statin use, aspirin use and the Charlson Comorbidity Index (using the updated version, categorized as 0, 1, 2 and ≥3) 7 . Insurance information was also obtained from the claims for the sensitivity analysis.
Statistical analysis
We first illustrated the flow chart of patient selection for the main analysis (the sitagliptin‐use and other DPP‐4 inhibitors‐use groups) and additional analysis (sulfonylurea and metformin groups).
Next, the patient characteristics were described for the main exposure groups. We then calculated the crude cancer incidence rates in each group. The cancer incidences were compared between groups using the bivariate Poisson regression analyses. Furthermore, Kaplan–Meier curves were generated to show the survival probabilities for the two groups, followed by a multivariable Cox proportional hazards model to adjust for covariates. To calculate the cancer incidence rates and carry out survival analyses, the patients were followed up until the month of the event and the end of the observation period, whichever came first.
We also repeated the analysis for each cancer type as a separate outcome variable. For the five major cancer types in Japan (i.e., stomach, colorectal, liver, lung and breast), two definitions of cancer incidence were used for one cancer type: (1) the first occurrence of the cancer diagnosis code, and (2) the combination of the cancer diagnosis code and medical practice code (surgery, chemotherapy, radiotherapy and palliative care). For the other cancer types, cancer incidence was calculated using the definition based on the diagnosis code only.
Various sensitivity analyses were carried out, and comparisons were repeated between the sitagliptin‐use group and the other DPP‐4 inhibitors‐use group. First, the exposure period was changed from 3 years to 1 or 5 years. Second, the analysis was restricted to patients aged ≤84 years at time 0 to remove the considerable competing risk of death among very old persons. Third, patients were censored at the month of the last medical contact, because this suggests unrecorded deaths due to other reasons or loss‐to‐follow up due to ID changes within individuals in the Japanese claims system. Fourth, insurance type, which was not included in the main analysis due to possible multicollinearity with age information, was included as a covariate in the Cox proportional hazards model.
Additionally, two other classes of medications, sulfonylurea and metformin, were introduced as exposure variables. The cancer incidences among the sulfonylurea‐use group and the metformin‐use group were calculated and compared with the sitagliptin‐use group, followed by the Cox proportional hazards analysis including all four groups (the sitagliptin‐use group, the other DPP‐4 inhibitor‐use group, the sulfonylurea‐use group and the metformin‐use group).
All statistical analyses were carried out using Stata 17.0 (StataCorp, College Station, TX, USA).
RESULTS
Patients' selection
The flowchart illustrating the patient selection process for the main analysis, comparing the sitagliptin group and the other DPP‐4 inhibitor group, is shown in Figure 1. Among patients who received at least one antidiabetic medication between 2012 and 2021 (n = 11,255,673), we identified those who had a regular prescription for sitagliptin or a DPP4 inhibitor other than sitagliptin for at least 3 years, and who could be observed for at least 6 months after time 0 (sitagliptin‐use group, n = 178,368; other DPP‐4 inhibitors‐use group, n = 240,980). After excluding patients with a time 0 of March 2021, those whose cancer care started on or before time 0 and those who were aged <20 years at time 0, the main analysis included 149,120 patients in the sitagliptin‐use group and 199,860 patients in the other DPP‐4 inhibitors‐use group. In sensitivity analyses, the exposure period was changed from 3 years to 1 or 5 years. For additional analysis, patients who initiated sulfonylurea or metformin and continued only the antidiabetic medication for at least 3 years were extracted (Figures S2 and S3).
Figure 1.
Flowchart of study participant selection in the main analysis. DPP‐4, dipeptidyl peptidase‐4; FY, fiscal year.
Patients' characteristics
Table S2 summarizes the characteristics of patients in the main exposure groups. There was no significant difference between the two groups in their characteristics. Out of a total of 348,980 patients, 52.4% were men. The largest age category was the 70–74‐year‐old group (19.2%), and 75.8% of the patients were aged ≥65 years.
Comparison of total cancer incidence between sitagliptin‐use and other DPP‐4 inhibitor‐use groups
Table 1 shows the total and sex‐ and age‐category‐stratified cancer incidence for the main exposure group. Within the sitagliptin‐use group (388,356 person‐years, average 2.6 years follow‐up [quartiles of 1.3, 2.7 and 3.9 years, 5 years in maximum]), 9,643 patients experienced cancer incidences (2,483.0 per 100,000 person‐years). Within the other DPP‐4 inhibitor‐use group (503,952 person‐years, average 2.5 years follow up [quartiles of 1.3, 2.5 and 3.8 years, 5 years in maximum]), 12,621 patients experienced cancer incidences (2,504.4 per 100,000 person‐years). The difference between the two groups was minimal, with an incidence rate ratio of 0.99 (95% confidence interval [CI] 0.97–1.02) according to the bivariate Poisson regression model. The analysis also showed no significant differences in cancer incidences between the groups when stratified by sex and age category, as demonstrated in Table S3. Figure 2a shows the Kaplan–Meier failure estimates for cancer incidence related to the two groups, showing little difference between them. In addition, the result of the Cox proportional hazards model (Table S4) showed that the use of sitagliptin, compared with other DPP‐4 inhibitors, was not associated with cancer incidence after controlling for covariates (hazard ratio [HR] 1.01, 95% CI 0.98–1.04). Figure 2b, depicting the plot of the failure function for the Cox proportional hazards model, shows a minimal difference between the two groups.
Table 1.
Total and sex‐ and age‐category‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or the other dipeptidyl peptidase‐4 inhibitors for 3 years in Japan from 2013 to 2018
| Sitagliptin‐use group | Other DPP‐4 inhibitor‐use group | P‐value † | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No. patients | Person‐years | No. events | Cancer incidence rate (per 100,000 person‐years) | No. patients | Person‐years | No. events | Cancer incidence rate (per 100,000 person‐years) | ||||||
| 95% CI | 95% CI | ||||||||||||
| LL | UL | LL | UL | ||||||||||
| Total | 149,120 | 388,356 | 9,643 | 2,483.0 | 2,434.0 | 2,533.1 | 199,860 | 503,952 | 12,621 | 2504.4 | 2461.1 | 2548.5 | 0.53 |
| Age category, years (male) | |||||||||||||
| 20–39 | 459 | 1,226 | <10 | – | – | – | 646 | 1,630 | <10 | – | – | – | – |
| 40–44 | 936 | 2,542 | ~10 | – | – | – | 1,298 | 3,447 | ~10 | – | – | – | – |
| 45–49 | 2,149 | 5,601 | 31 | 553.5 | 389.3 | 787.0 | 3,187 | 7,999 | 50 | 625.1 | 473.7 | 824.7 | 0.60 |
| 50–54 | 4,075 | 10,769 | 85 | 789.3 | 638.1 | 976.3 | 5,592 | 14,309 | 135 | 943.4 | 797.0 | 1116.8 | 0.20 |
| 55–59 | 6,596 | 17,335 | 232 | 1,338.3 | 1,176.7 | 1,522.1 | 8,592 | 21,792 | 253 | 1161.0 | 1026.4 | 1313.2 | 0.12 |
| 60–64 | 9,742 | 25,784 | 462 | 1,791.8 | 1,635.6 | 1,962.9 | 12,661 | 32,673 | 594 | 1818.0 | 1677.5 | 1970.3 | 0.82 |
| 65–69 | 14,689 | 39,491 | 1,080 | 2,734.8 | 2,576.5 | 2,902.9 | 19,343 | 50,351 | 1,346 | 2673.2 | 2534.2 | 2819.9 | 0.58 |
| 70–74 | 14,689 | 36,258 | 1,308 | 3,607.5 | 3,417.2 | 3,808.4 | 19,979 | 47,101 | 1,772 | 3762.2 | 3591.0 | 3941.5 | 0.25 |
| 75–79 | 12,008 | 29,786 | 1,312 | 4,404.8 | 4,172.7 | 4,649.7 | 17,133 | 41,775 | 1,790 | 4284.9 | 4090.9 | 4488.0 | 0.45 |
| 80–84 | 7,608 | 19,075 | 904 | 4,739.2 | 4,440.1 | 5,058.4 | 11,086 | 26,961 | 1,191 | 4417.6 | 4173.7 | 4675.7 | 0.11 |
| 85 | 4,273 | 10,643 | 495 | 4,651.0 | 4,258.8 | 5,079.3 | 6,295 | 15,204 | 726 | 4775.0 | 4440.0 | 5135.2 | 0.65 |
| Age category, years (female) | |||||||||||||
| 20–39 | 189 | 491 | <10 | – | – | – | 287 | 735 | <10 | – | – | – | – |
| 40–44 | 333 | 903 | <10 | – | – | – | 440 | 1,117 | <10 | – | – | – | – |
| 45–49 | 788 | 1984 | 29 | 1,461.6 | 1,015.7 | 2,103.3 | 1,092 | 2,746 | 32 | 1165.3 | 824.1 | 1647.9 | 0.38 |
| 50–54 | 1,487 | 3,863 | 33 | 854.3 | 607.3 | 1,201.7 | 2,119 | 5,355 | 68 | 1269.8 | 1001.2 | 1610.5 | 0.06 |
| 55–59 | 3,409 | 9,092 | 115 | 1,264.9 | 1,053.6 | 1,518.5 | 4,382 | 11,344 | 129 | 1137.1 | 956.9 | 1351.3 | 0.41 |
| 60–64 | 6,151 | 16,771 | 237 | 1,413.2 | 1,244.2 | 1,605.0 | 7,795 | 20,415 | 293 | 1435.2 | 1279.9 | 1609.3 | 0.86 |
| 65–69 | 11,615 | 31,995 | 488 | 1,525.3 | 1,395.8 | 1,666.8 | 14,833 | 39,659 | 652 | 1644.0 | 1522.5 | 1775.2 | 0.21 |
| 70–74 | 14,245 | 36,545 | 731 | 2,000.3 | 1,860.4 | 2,150.7 | 18,244 | 45,284 | 882 | 1947.7 | 1823.3 | 2080.6 | 0.59 |
| 75–79 | 14,022 | 36,297 | 778 | 2,143.4 | 1,998.0 | 2,299.4 | 18,184 | 45,636 | 962 | 2108.0 | 1978.9 | 2245.5 | 0.73 |
| 80–84 | 10,346 | 27,303 | 650 | 2,380.7 | 2,204.5 | 2,570.9 | 14,108 | 36,128 | 876 | 2424.7 | 2269.3 | 2590.7 | 0.72 |
| 85 | 9,311 | 24,604 | 648 | 2,633.7 | 2,438.5 | 2,844.5 | 12,564 | 32,290 | 843 | 2610.7 | 2440.3 | 2793.0 | 0.87 |
The sitagliptin‐use group had initiated and continued sitagliptin for 3 years before time 0. However, the other dipeptidyl peptidase‐4 (DPP‐4) inhibitor‐use group had initiated and continued DPP‐4 inhibitors other than sitagliptin for 3 years before time 0, with the allowance for drug changes within the DPP‐4 inhibitor class. Cancer incidence was detected based on “the state date of care” for cancers listed in the diagnosis information in the claims. Censoring was applied either at the time of the cancer incidence event or at the end of the observation period (March 2021). Numbers with counts <10 were masked in compliance with regulatory guidelines. CI, confidence interval; DPP‐4, dipeptidyl peptidase‐4; LL, lower limit; UL, upper limit.
Comparison between the sitagliptin‐use group and the other DPP‐4 inhibitor‐use group using the bivariate Poisson regression model.
Figure 2.
Cumulative cancer incidence in sitagliptin versus other dipeptidyl peptidase‐4 inhibitors (DPP‐4i) in 3‐year consecutive users in Japan. (a) Kaplan–Meier failure estimates for cancer incidence. Incidence rate ratio of 0.99 (95% confidence interval 0.97–1.02), P = 0.53, for the bivariate Poisson regression model. (b) Plots showing the failure function for the Cox proportional hazards model, adjusted for sex, age category, interaction terms of sex and age category, prefecture, statin use, aspirin use, and the Charlson Comorbidity Index. Hazard ratio of 1.01 (95% confidence interval 0.98–1.04), P = 0.52, for the Cox proportional hazards model.
Comparison of the incidence of each cancer type between sitagliptin‐use and other DPP‐4 inhibitor‐use groups
In the analyses for each cancer type as a separate outcome variable, the use of DPP‐4 inhibitors was not strongly associated with cancer incidence for the five major cancer types of Japan, regardless of whether the diagnosis code or the combination of diagnosis code and medical practice code was used (Table 2). The largest HR was observed for lung cancer, defined by the diagnosis and medical practice codes (HR 1.05, 95% CI 0.96–1.16). In addition, Table S5 presents the cancer incidence (defined by diagnosis code) for each cancer type other than the five major cancers in Japan. Although the use of sitagliptin, compared with other DPP‐4 inhibitors, was associated with an increased incidence of larynx cancer and a decreased incidence of cervical cancer, each cancer incidence was scattered around the point estimate of 1.
Table 2.
Comparison of the incidence of each cancer (Japan's five major cancers) associated with sitagliptin use and other dipeptidyl peptidase‐4 inhibitors
| Sitagliptin‐use group | Other DPP‐4 inhibitor‐use group | Bivariate Poisson regression model | Multivariable Cox proportional hazards model | |||
|---|---|---|---|---|---|---|
| Cancer incidence rate (per 100,000 person‐years) (95% CI) | Cancer incidence rate (per 100,000 person‐years) (95% CI) | Incidence rate ratio (95% CI) | P‐value | Hazard ratio (95% CI) | P‐value | |
| Stomach cancer (diagnosis) | 356.4 (338.4–375.3) | 354.0 (338.2–370.5) | 1.01 (0.94–1.08) | 0.85 | 1.04 (0.97–1.11) † | 0.30 |
| Stomach cancer (diagnosis and medical practice) | 243.4 (228.7–259.1) | 240.6 (227.7–254.3) | 1.01 (0.93–1.10) | 0.78 | 1.04 (0.96–1.13) † | 0.35 |
| Colorectal cancer (diagnosis) | 476.8 (456.0–498.7) | 485.5 (466.9–504.7) | 0.98 (0.93–1.04) | 0.55 | 0.99 (0.93–1.05) † | 0.75 |
| Colorectal cancer (diagnosis and medical practice) | 345.2 (327.6–363.9) | 344.8 (329.2–361.0) | 1.00 (0.93–1.07) | 0.97 | 1.00 (0.94–1.08) † | 0.90 |
| Liver cancer (diagnosis) | 133.5 (122.7–145.2) | 132.3 (122.8–142.5) | 1.01 (0.90–1.13) | 0.87 | 1.05 (0.94–1.18) ‡ | 0.40 |
| Liver cancer (diagnosis and medical practice) | 76.8 (68.7–85.8) | 76.5 (69.4–84.4) | 1.00 (0.86–1.17) | 0.96 | 1.05 (0.90–1.21) ‡ | 0.55 |
| Lung cancer (diagnosis) | 336.3 (318.9–354.7) | 336.5 (320.9–352.7) | 1.00 (0.93–1.07) | 0.98 | 1.02 (0.95–1.10) ‡ | 0.57 |
| Lung cancer (diagnosis and medical practice) | 204.1 (190.7–218.6) | 193.6 (182.0–205.9) | 1.05 (0.96–1.16) | 0.26 | 1.07 (0.97–1.17) ‡ | 0.16 |
| Breast cancer (diagnosis) | 304.5 (281.0–330.0) | 303.4 (282.5–325.9) | 1.00 (0.90–1.12) § | 0.95 | 1.00 (0.90–1.12) ¶ | 0.97 |
| Breast cancer (diagnosis and medical practice) | 251.5 (230.2–274.7) | 251.8 (232.8–272.3) | 1.00 (0.89–1.13) § | 0.99 | 0.99 (0.88–1.12) ¶ | 0.89 |
Cancer types other than those listed above are described in Table S6. CI, confidence interval; DPP‐4, dipeptidyl peptidase.
Restricted to age ≥40 years.
Restricted to age ≥50 years.
Restricted to women only.
Restricted to women and age ≥40 years.
Sensitivity analyses
Table 3 summarizes the results of sensitivity analysis, delineating the association between cancer incidence in the sitagliptin‐use group and the other DPP‐4 inhibitors‐use group. First, changing the exposure period from 3 years to 1 or 5 years hardly moved the point estimate of the incidence rate ratio or HR. Tables S6 and S7 present the total and sex‐ and age‐category‐stratified cancer incidence in these analyses. Second, restricting the analysis to patients aged ≤84 years did not alter the results in either the bivariate Poisson regression model or the multivariable Cox proportional hazards model. Third, the effect measures remained robust when adding another censor timing of “the month of the most recent medical contact for each individual,” related to cancer incidence (Table S8). Fourth, including insurance type as a covariate in the Cox proportional hazards model had minimal impact on the main results.
Table 3.
Comparison of the sensitivity analyses for cancer incidence associated with sitagliptin use and other dipeptidyl peptidase‐4 inhibitors use
| Sitagliptin‐use group | Other DPP‐4 inhibitor‐use group | Bivariate Poisson regression model | Multivariable Cox proportional hazards model | |||
|---|---|---|---|---|---|---|
| Cancer incidence rate (per 100,000 person‐years) (95% CI) | Cancer incidence rate (per 100,000 person‐years) (95% CI) | Incidence rate ratio (95% CI) | P‐value | Hazard ratio (95% CI) | P‐value | |
| (Main analysis) Continuous use period: 3 years | 2,483.0 (2,434.0–2,533.1) | 2,504.4 (2,461.1–2,548.5) | 0.99 (0.97–1.02) | 0.53 | 1.01 (0.98–1.04) | 0.52 |
| Continuous use period: 1 year | 2,365.0 (2,339.2–2,391.0) | 2,367.1 (2,344.5–2,389.9) | 1.00 (0.98–1.01) | 0.9 | 1.01 (1.00–1.03) | 0.09 |
| Continuous use period: 5 years | 2,508.7 (2,407.1–2,614.5) | 2,608.1 (2,515.4–2,704.2) | 0.96 (0.91–1.02) | 0.17 | 0.98 (0.92–1.03) | 0.4 |
| Restricted to patients aged ≤84 years | 2,407.2 (2,356.6–2,458.9) | 2,421.3 (2,376.5–2,466.8) | 0.99 (0.97–1.02) | 0.69 | 1.01 (0.98–1.04) | 0.47 |
| Censoring was applied at the earliest of the following events: cancer incidence event, last claim included in the dataset or end of the observation period (March 2021) | 2,550.7 (2,500.3–2,602.1) | 2,583.0 (2,538.3–2,628.5) | 0.99 (0.96–1.02) | 0.41 | 1.00 (0.98–1.03) | 0.78 |
| Cox proportional hazards model, including insurance type as a covariate in the model | 1.01 (0.98–1.04) | 0.51 | ||||
CI, confidence interval; DPP‐4, dipeptidyl peptidase.
Comparison of the total incidence of cancer between the sulfonylurea‐use, metformin‐use groups and the sitagliptin‐use group
In the additional analysis, two other medication classes (sulfonylurea and metformin) were included as exposures. It was found that the sulfonylurea‐use group had a higher cancer incidence rate. In comparison, the metformin‐use group had a lower cancer incidence rate compared with the sitagliptin‐use group. Table S9 presents the sex‐ and age‐category‐stratified cancer incidence among study participants who had initiated and continued sulfonylureas or metformin for 3 years. The difference remained after adjusting for covariates in the Cox proportional hazards model (in comparison with sitagliptin, HR for sulfonylurea 1.06, 95% CI 1.00–1.12; HR for metformin 0.94, 95% CI 0.90–0.98).
DISCUSSION
Based on our investigation in the present observational study, there is no evidence to suggest that the use of sitagliptin (found to contain NTTP, according to the FDA) is associated with an increased cancer incidence within the observation period. The use of a comprehensive claims database covering >120 million people in Japan enabled us to compare patients who initiated antidiabetic medication with sitagliptin or other DPP‐4 inhibitors and continued the medication exclusively for a certain period in terms of cancer incidence risk. Various sensitivity analyses were carried out to ensure the robustness of the study results. To the best of our knowledge, this is the first study to investigate the relationship between the use of NTTP‐contaminated sitagliptin, compared with other DPP‐4 inhibitors, and cancer incidence.
In contrast to the lack of apparent difference between the sitagliptin‐use group and the other DPP‐4 inhibitor‐use group, we observed noticeable differences when comparing across medications classes (sitagliptin vs sulfonylurea and sitagliptin vs metformin). From a pharmacological perspective, insulin affects tumor growth 8 , and sulfonylurea promotes insulin secretion, whereas metformin alleviates hyperinsulinemia by improving insulin resistance. For comparison between sulfonylurea and DPP‐4 inhibitors, both of which facilitate insulin secretion, DPP‐4 inhibitors prevented cancer by promoting lymphocyte trafficking and raising tumor immunity 9 . The findings of the present study are consistent with these pharmacological insights and some earlier epidemiological findings 10 , 11 , 12 , 13 . However, cohort studies have struggled to fully isolate the true effects of antidiabetic drugs on carcinogenesis from a variety of biases, leading to the ongoing controversy surrounding the association between the class of antidiabetic medication and the risk of cancer incidence 14 . Although some biases, such as prevalent‐user bias and detection bias, were overcome by limiting the analysis to new users who used only one type of antidiabetic medication, the observed difference in cancer incidence among DPP‐4 inhibitor users, sulfonylurea users and metformin users should be interpreted with caution due to possible confounding. The observed higher cancer risk for sulfonylurea users and lower risk for metformin users aligned with a previous study on cardiovascular risk associated with antidiabetic medication 15 . Regardless, it is noteworthy that there was no difference when comparing NTTP‐contaminated sitagliptin and other DPP‐4 inhibitors within the same class, although a noticeable difference was observed when comparing across the classes (sulfonylureas and metformin to sitagliptin).
Each type of cancer was also considered an outcome in the present study, allowing us to investigate the heterogeneity by cancer type. For most cancer types, the incidence rate ratios in bivariate Poisson regression models and HRs in multivariable Cox proportional hazards models were approximately 1. However, we observed an increased risk of larynx cancer incidence (HR 1.58, 95% CI 1.11–2.24) and a decreased risk of cervical cancer incidence (HR 0.60, 95% CI 0.41–0.86). To the best of our knowledge, there have been no reports on organ specificity for NTTP carcinogenesis. These associations could be due to chance; however, further investigation is necessary.
The present investigation was motivated by pharmacoepidemiological studies that utilized big data, including claims data, to examine the adverse outcomes of drugs 16 , 17 , 18 , 19 . None of these studies found any evidence of an increased risk of cancer. One possible explanation for this is that the original acceptable intake limit was set quite strictly, such that any additional risk would only occur after a lifetime of exposure to the contaminated substance. Detecting such a difference using the data from only a few years of observation might have been a serious disadvantage. The use of a large sample size in claims data is particularly advantageous for studying the occurrence of rare events or adverse outcomes.
The present study had several limitations. First, as this was an observational study based on claims data, the causality remains to be elucidated. Notably, we were unable to control for some potential confounders, such as smoking status and body mass index. However, there is no reason to believe that sitagliptin users would smoke more frequently and weigh more than other DPP‐4 inhibitor users. Second, the average follow‐up time of the main analysis was approximately 3 years (5 years at maximum), with an additional 3‐year exposure period. Furthermore, the present real‐world data analysis did not adopt active cancer screening. We believe this is the largest database with which we can tackle the research question and the cancer screening rate in Japan is high (40%–50%) 20 ; however, the exposure and follow‐up period would be insufficient to conclude the null effect of nitrosamine impurity. Further investigations with longer follow‐up periods are warranted. Third, the validity of the diagnoses of claims data for identifying cancer incidence was reported to be good 6 , but not perfect. Even the nondifferential misclassification of the outcome could cause a bias toward the null. However, after identifying the five major cancers using the diagnosis codes and medical practice codes, we found no difference in their incidence rates between the sitagliptin‐use and other DPP‐4 inhibitor‐use groups. Fourth, metformin was reported to be contaminated with another type of nitrosamine, N‐nitrosodimethylamine 21 . Due to the small sample size of those who only used contaminated metformin, we could not sort out the effect of contaminated metformin. Fifth, we did not adopt a noninferiority study design; therefore, higher P‐values should not be interpreted as evidence of no difference between the groups. Finally, due to the lack of ledger information in the NDB, it is impossible to identify patient death perfectly. Furthermore, given that some individuals might have two or more IDs, it is challenging to ascertain whether someone who disappeared from the NDB at a specific time died, became untraceable or ceased to seek medical care. Linking NDB and ledger information to accurately determine censorship would be ideal. To overcome this limitation, we carried out sensitivity analysis by additionally censoring patients at the month of the last medical contact and obtained similar results.
In conclusion, the present cohort study investigated the association between the use of NTTP‐contaminated sitagliptin and cancer incidence using Japan's national claims data in response to concerns raised by the FDA regarding NTTP contamination of sitagliptin. Our finding supports no evidence to suggest an increased cancer incidence among patients prescribed NTTP‐contaminated sitagliptin for 3 years compared with those prescribed other DPP‐4 inhibitors.
DISCLOSURE
MO has received grants or contracts from Abbott Japan LLC, Astellas Pharma Inc., Eli Lilly Japan K.K., Kyowa Kirin Co., MSD K.K., Nippon Boehringer Ingelheim Co., Ltd., Novo Nordisk Japan Pharma Ltd. and Sanofi K.K.; and payment or honoraria from Abbott Japan LLC, Astellas Pharma Inc., Bayer Holding Ltd., Daiichi Sankyo Company, Limited, Eli Lilly Japan K.K., Kowa Company, Ltd., Mitsubishi Tanabe Pharma Corporation, MSD K.K., Nippon Boehringer Ingelheim Co., Ltd., Novartis Pharma K.K., Novo Nordisk Japan Pharma Ltd., Ono Pharmaceutical Co, Ltd., Sanofi K.K., Sanwa Kagaku Kenkyusho, Co., Ltd., Sumitomo Pharma, Co., Ltd., Takeda Pharmaceutical Co., Ltd. and Teijin Pharma Limited. KU received grants or contracts from Sumitomo Pharma Co., Ltd., Nippon Boehringer Ingelheim Co., Ltd., Eli Lilly Japan K.K., Novo Nordisk Japan Pharma Ltd. and Sanofi K.K.; payment or honoraria from Sumitomo Pharma Co., Ltd., Bayer Yakuhin, Ltd., Novo Nordisk Japan Pharma Ltd., Nippon Boehringer Ingelheim Co., Ltd., Taisho Pharmaceutical Co., Ltd., Kowa Company, Ltd., Eli Lilly Japan K.K., Daiichi Sankyo Company, Limited. and Abbott Japan LLC.; and participated in a data safety monitoring board or advisory board for Abbott Japan LLC, Kyowa Kirin Co., Ltd., Bayer Yakuhin Ltd., Terumo Corporation, AstraZeneca, Eli Lilly Japan K.K., Sumitomo Pharma, Co., Ltd., Mitsubishi Tanabe Pharma Corporation and Novo Nordisk Japan Pharma Ltd. There are no other relationships or activities that could have influenced the submitted work. KU is an Editorial Board member of the Journal of Diabetes Investigation and a co‐author of this article. To minimize bias, they were excluded from all editorial decision‐making related to the acceptance of this article for publication.
Approval of the research protocol: Our study proposal was approved by the Institutional Review Board of the National Center for Global Health and Medicine (NCGM‐G‐004081‐06).
Informed consent: N/A.
Approval date of registry and the registration no. of the study/trial: N/A.
Animal studies: N/A.
REPORTING CHECKLIST
We included the checklist from the Reporting of studies Conducted using Observational Routinely‐collected Data (RECORD) statement, which is an extension of the STROBE guidelines (Table S10). During the preparation of this work, the authors used ChatGPT for the purpose of manuscript writing. Following the use of this tool/service, the authors formally reviewed the content for accuracy and performed the necessary editing. The authors take full responsibility for all the content provided in this publication.
Supporting information
Data S1. Supplemental Methods.
Table S1. Definition of each variable.
Table S2. Characteristics of study participants who had initiated and continued sitagliptin or other DPP‐4 inhibitors for 3 years in Japan from 2013 to 2018.
Table S3. Characteristics‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or other DPP‐4 inhibitors for 3 years in Japan from 2013 to 2018.
Table S4. Factors associated with cancer incidence among DPP‐4 inhibitor users: Cox proportional hazards model (n = 321,812).
Table S5. Sensitivity analyses of cancer incidence associated with sitagliptin use compared with other DPP‐4 inhibitors: analyses by cancer location for cancer types other than the five major types.
Table S6. Gender‐ and age‐category‐stratified cancer incidence among study participants who initiated and continued sitagliptin or other DPP‐4 inhibitors for 1 year in Japan from 2013 to 2020.
Table S7. Gender‐ and age‐category‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or the other DPP‐4 inhibitors for 5 years in Japan from 2013 to 2016.
Table S8. Gender and age‐category‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or the other DPP‐4 inhibitors for 3 years in Japan from 2013 to 2018.
Table S9. Gender and age‐category‐stratified cancer incidence among study participants who initiated and continued sulfonylurea or metformin for 3 years in Japan from 2013 to 2018.
Table S10. The RECORD statement—checklist of items, extended from the STROBE statement that should be reported in observational studies using routinely collected health data.
Figure S1. Longitudinal study design.
Figure S2. Flowchart of study participant selection for additional analyses: sulfonylurea‐use group.
Figure S3. Flowchart of study participant selection for additional analyses: metformin‐use group.
ACKNOWLEDGMENTS
The authors thank all members of the Research Project for the Establishment of an NDB Research System for Health Policy and Other Purposes through the 6NC Collaboration. The authors also appreciate Denno Labo Corporation for their contribution to data processing. The authors also thank Enago (https://www.enago.jp) for their English language editing. This work was supported by the Japan Health Research Promotion Bureau (grant number 2019‐[1]‐3). The funding agency was not involved in study design, collection, analysis, interpretation of data, writing the report or any restrictions regarding submitting the report for publication.
DATA AVAILABILITY STATEMENT
Data supporting the findings of this study are available from the Ministry of Health, Labor and Welfare. However, restrictions apply to the availability of these data, which were used under license for the current study and, therefore, are not publicly available. However, data are available from the Ministry upon reasonable request and with permission from the Ministry.
REFERENCES
- 1. Fang M, Wang D, Coresh J, et al. Trends in diabetes treatment and control in U.S. adults, 1999‐2018. N Engl J Med 2021; 384: 2219–2228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Bouchi R, Sugiyama T, Goto A, I Imai K, Ihana‐Sugiyama N, Ohsugi M, Yamauchi T, Kadowaki T, Ueki K Retrospective nationwide study on the trends in first‐line antidiabetic medication for patients with type 2 diabetes in Japan. J Diabetes Investig 2022;13:280–291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. U.S. Food and Drug Administration . FDA works to avoid shortage of sitagliptin following detection of nitrosamine impurity. 2022. Available from: https://www.fda.gov/drugs/drug‐safety‐and‐availability/fda‐works‐avoid‐shortage‐sitagliptin‐following‐detection‐nitrosamine‐impurity Accessed May 30, 2024.
- 4. U.S. Food and Drug Administration . Control of nitrosamine Impurities in Human Drugs–Guidance for Industry. 2021. Available from: https://www.fda.gov/media/141720/download Accessed May 30, 2024.
- 5. Kato G. History of the secondary use of national database of health insurance claims and specific health checkups of Japan (NDB). Trans Jpn Soc Med Biol Eng 2017; 55: 143–150. [Google Scholar]
- 6. Ihira H, Goto A, Yamagishi K, et al. Validity of claims data for identifying cancer incidence in the Japan public health center‐based prospective study for the next generation. Pharmacoepidemiol Drug Saf 2022; 31: 972–982. [DOI] [PubMed] [Google Scholar]
- 7. Quan H, Li B, Couris CM, et al. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol 2011; 173: 676–682. [DOI] [PubMed] [Google Scholar]
- 8. Pollak M. The insulin and insulin‐like growth factor receptor family in neoplasia: An update. Nat Rev Cancer 2012; 12: 159–169. [DOI] [PubMed] [Google Scholar]
- 9. Barreira da Silva R, Laird ME, Yatim N, et al. Dipeptidylpeptidase 4 inhibition enhances lymphocyte trafficking, improving both naturally occurring tumor immunity and immunotherapy. Nat Immunol 2015; 16: 850–858. [DOI] [PubMed] [Google Scholar]
- 10. Evans JMM, Donnelly LA, Emslie‐Smith AM, et al. Metformin and reduced risk of cancer in diabetic patients. BMJ 2005; 330: 1304–1305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Gandini S, Puntoni M, Heckman‐Stoddard BM, D Dunn BK, Ford L, DeCensi A, Szabo E Metformin and cancer risk and mortality: A systematic review and meta‐analysis taking into account biases and confounders. Cancer Prev Res 2014; 7: 867–885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Wu L, Zhu J, Prokop LJ, et al. Pharmacologic therapy of diabetes and overall cancer risk and mortality: A meta‐analysis of 265 studies. Sci Rep 2015; 5: 10147. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Pradhan R, Yu OHY, Platt RW, et al. Dipeptidyl peptidase‐4 inhibitors and the risk of skin cancer among patients with type 2 diabetes: A UK population‐based cohort study. BMJ Open Diabetes Res Care 2023; 11: e003550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Klil‐Drori AJ, Azoulay L, Pollak MN. Cancer, obesity, diabetes, and antidiabetic drugs: Is the fog clearing? Nat Rev Clin Oncol 2017; 14: 85–99. [DOI] [PubMed] [Google Scholar]
- 15. Komamine M, Kajiyama K, Ishiguro C, et al. Cardiovascular risks associated with dipeptidyl peptidase‐4 inhibitors monotherapy compared with other antidiabetes drugs in the Japanese population: A nationwide cohort study. Pharmacoepidemiol Drug Saf 2019; 28: 1166–1174. [DOI] [PubMed] [Google Scholar]
- 16. Iwagami M, Kumazawa R, Miyamoto Y, et al. Risk of cancer in association with ranitidine and nizatidine vs. other h2 blockers: Analysis of the Japan medical data center claims database 2005‐2018. Drug Saf 2021; 44: 361–371. [DOI] [PubMed] [Google Scholar]
- 17. Pottegård A, Kristensen KB, Ernst MT, et al. Use of N‐nitrosodimethylamine (NDMA) contaminated valsartan products and risk of cancer: Danish nationwide cohort study. BMJ 2018; 362: k3851. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Yoon HJ, Kim JH, Seo GH, et al. Risk of cancer following the use of N‐Nitrosodimethylamine (NDMA) contaminated ranitidine products: A nationwide cohort study in South Korea. J Clin Med 2021; 10: 153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Kantor ED, O'Connell K, Du M, et al. Ranitidine use and cancer risk: Results from UK biobank. Gastroenterology 2021; 160: 1856–1859.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Ministry of Health, Labour and Welfare . Summary Report of Comprehensive Survey of Living Conditions. 2020. Available from: https://www.mhlw.go.jp/english/database/db‐hss/dl/report_gaikyo_2019.pdf Accessed May 30, 2024.
- 21. U.S. Food and Drug Administration . Questions and Answers: NDMA impurities in metformin products. 2020. Available from: https://www.fda.gov/drugs/drug‐safety‐and‐availability/questions‐and‐answers‐ndma‐impurities‐metformin‐products Accessed May 30, 2024.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Supplemental Methods.
Table S1. Definition of each variable.
Table S2. Characteristics of study participants who had initiated and continued sitagliptin or other DPP‐4 inhibitors for 3 years in Japan from 2013 to 2018.
Table S3. Characteristics‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or other DPP‐4 inhibitors for 3 years in Japan from 2013 to 2018.
Table S4. Factors associated with cancer incidence among DPP‐4 inhibitor users: Cox proportional hazards model (n = 321,812).
Table S5. Sensitivity analyses of cancer incidence associated with sitagliptin use compared with other DPP‐4 inhibitors: analyses by cancer location for cancer types other than the five major types.
Table S6. Gender‐ and age‐category‐stratified cancer incidence among study participants who initiated and continued sitagliptin or other DPP‐4 inhibitors for 1 year in Japan from 2013 to 2020.
Table S7. Gender‐ and age‐category‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or the other DPP‐4 inhibitors for 5 years in Japan from 2013 to 2016.
Table S8. Gender and age‐category‐stratified cancer incidence among study participants who had initiated and continued sitagliptin or the other DPP‐4 inhibitors for 3 years in Japan from 2013 to 2018.
Table S9. Gender and age‐category‐stratified cancer incidence among study participants who initiated and continued sulfonylurea or metformin for 3 years in Japan from 2013 to 2018.
Table S10. The RECORD statement—checklist of items, extended from the STROBE statement that should be reported in observational studies using routinely collected health data.
Figure S1. Longitudinal study design.
Figure S2. Flowchart of study participant selection for additional analyses: sulfonylurea‐use group.
Figure S3. Flowchart of study participant selection for additional analyses: metformin‐use group.
Data Availability Statement
Data supporting the findings of this study are available from the Ministry of Health, Labor and Welfare. However, restrictions apply to the availability of these data, which were used under license for the current study and, therefore, are not publicly available. However, data are available from the Ministry upon reasonable request and with permission from the Ministry.