Comparison of the Effectiveness of Vedolizumab and Ustekinumab in Crohn’s Disease Patients Who Failed Anti-tumor Necrosis Factor-α Treatment in Japan: An Observational Study Utilizing Claims Database

Abstract

This study aimed to investigate whether the approved sequence of vedolizumab and ustekinumab impacts the results of previous observational studies conducted in the European Union (EU), comparing the effectiveness of these drugs in Crohn’s disease (CD) patients who failed anti-tumor necrosis factor-α (TNFα) treatment. We conducted this study in Japan, where the approved sequence of drugs is different from that of the EU. We extracted 256 patients diagnosed with CD, who had a history of anti-TNFα treatment and were prescribed either vedolizumab or ustekinumab, from JMDC claims database. The patients’ backgrounds were adjusted by inverse probability of treatment weighting using propensity score. The primary outcome was treatment persistence. Secondary outcomes were a steroid-free period, time to hospitalization, and time to CD-related surgery. The hazard ratios (HR) for survival times were estimated using the Cox proportional hazard model. The treatment persistence (primary endpoint) was significantly longer for ustekinumab than vedolizumab (HR, 0.32; 95% confidence interval (CI), 0.15–0.72). The results of the secondary endpoints were as follows: steroid-free period (HR, 0.38; 95% CI, 0.10–1.48), time to hospitalization (HR, 1.07; 95% CI, 0.60–1.91), or time to CD-related surgery (HR, 0.33; 95% CI, 0.11–0.97). There were no outcomes indicating the superiority of vedolizumab. Our findings suggest that ustekinumab is a more effective treatment option than vedolizumab for CD patients who failed to anti-TNFα treatment, and this finding remains consistent across both Japan and the EU.

INTRODUCTION

Crohn’s disease (CD) is a type of inflammatory bowel disease that is characterized by long-term inflammation in any part of the digestive system. It is a progressively destructive condition that is becoming more common globally.¹⁾ Anti-tumor necrosis factor-α (TNFα) antibodies, such as infliximab and adalimumab are effective in treating inflammatory bowel diseases, leading to improvements in patient’s QOL and a decrease in surgeries and hospitalizations. However, a substantial subset of patients exhibit unresponsiveness to the initial treatment or experience a decline in response over time.^2,3) For patients who fail to anti-TNFα treatment, other biological agents such as vedolizumab and ustekinumab are recommended.⁴⁾ Vedolizumab is a monoclonal antibody that blocks the α4β7 integrin, resulting in gut-selective anti-inflammatory activity. Ustekinumab is a monoclonal antibody that binds to the p40 subunit shared by the pro-inflammatory interleukins 12 and 23.⁴⁾

Observational studies conducted in France, the Netherlands, England, and the United Kingdom comparing vedolizumab and ustekinumab for the anti-TNFα treatment failure patients support the efficacy of ustekinumab over vedolizumab.^5–9) However, there have been no randomized clinical trials comparing these two drugs. Furthermore, the order in which the drugs were approved for use in the European Union (EU)—vedolizumab before ustekinumab—¹⁰⁾ may have introduced unmeasured confounding factors that could have affected the results of these observational studies.

In Japan, contrary to the EU, ustekinumab was approved on 17 February, 2017 before vedolizumab which obtained approval on 22 May, 2019 for CD.^11,12) Therefore, we aimed to compare the effectiveness of vedolizumab and ustekinumab in patients who failed to anti-TNFα treatment using Japanese data to determine if the order of drug approval had an impact on the findings of previous observational studies.

MATERIALS AND METHODS

We conducted a longitudinal database cohort study comparing treatment persistence, steroid-free period, time to hospitalization, and time to CD-related surgery for CD patients in users of vedolizumab with users of ustekinumab who failed to anti-TNFα treatment. A study design diagram was presented in Fig. 1. A list of all codes used to define exposure, outcome, and covariates in this study can be found in Supplementary Material 1. The diagnosis of CD (inclusion criteria) and covariates, including CD diagnosis age, CD duration, perianal disease, extraintestinal manifestations, psoriasis, and pouchitis, were confirmed or calculated using disease records without suspicion flags. However, when checking whether the patient initiated anti-TNFα treatment before being diagnosed with CD or not, suspicion flag for CD was not taken into consideration.

Fig. 1. Study Design Diagram

CD, Crohn’s disease.

Data Source

We used the commercially available administrative claims database from JMDC Inc. (Tokyo, Japan), which is compiled from multiple health insurance societies in Japan and includes outpatient, inpatient, and pharmacy claims since January 2005.¹³⁾ All data are anonymized so that individuals cannot be identified. Chronological tracking of data is readily feasible, even when an individual has received medical care at multiple institutions. As of September 2021, the end of the data period for this study, the database includes approximately 13.7 million insured individuals. JMDC database does not include data on individuals over the age of 75, as health insurance societies in Japan only cover employees of medium to large companies and their dependents under the age of 75. While JMDC database includes information on the birth month and last observation month, it does not contain accurate date data. Therefore, we assigned a hypothetical date of the 15th of the month to calculate the age or observation period.

Study Population

Eligible patients were diagnosed with CD, prescribed at least one anti-TNFα antibody following their CD diagnosis, and initiated treatment with vedolizumab or ustekinumab after May 2019. The prescription of vedolizumab or ustekinumab had to occur after anti-TNFα treatment. Patients were separated into vedolizumab or ustekinumab groups on the index date, defined as the earliest date of vedolizumab or ustekinumab prescription. Exclusion criteria included: a look-back period of fewer than 365 d; being under 18 years of age on the index date; having pouchitis within the look-back period; and having an index date within the final observation month for each patient.

Outcome and Follow-Up

The primary endpoint of this study was treatment persistence with vedolizumab or ustekinumab. Vedolizumab or ustekinumab was considered discontinued if more than 120 d had passed since administration. Secondary endpoints included a steroid-free period, time to hospitalization, and time to CD-related surgery. To assess the steroid-free period, we excluded patients prescribed steroids within 30 d of the index date. For the assessment of time to hospitalization, we excluded patients already hospitalized at the index date. Regarding the assessment of time to CD-related surgery, we excluded patients who had undergone CD-related surgery within the look-back period. The study participants were followed up until the occurrence of the outcome, withdrawal from JMDC database, or the end of the data period (September 2021).

Covariates

Covariates that may affect the outcomes, including sex, age on the index date, prior prescription of infliximab or adalimumab, use of immunosuppressants (azathioprine, methotrexate, 6-mercaptopurine) or steroids within 30 d of the index date, CD-related surgery, perianal disease, extraintestinal manifestations, psoriasis within the look-back period, age at CD diagnosis (under 41 vs. over 41), and CD duration, selected based on previous studies,^5–9) were adjusted for in the analysis. Disease severity data such as Crohn’s disease activity index, Harvey–Bradshaw index, and C-reactive protein (CRP) were not available in JMDC database.

Statistical Analysis

To control for differences in baseline characteristics, we conducted analyses using inverse probability of treatment weighting (IPTW) based on propensity score (PS). The PS was calculated using a logistic regression model for receipt of ustekinumab treatment, incorporating the aforementioned covariates. We compared the baseline characteristics of patients using standardized mean differences (SMD) between the two groups. The probability of survival (treatment persistence, steroid-free period, time to hospitalization, and time to CD-related surgery) was estimated with the Kaplan–Meier (KM) method, and hazard ratios (HR) were estimated using the Cox proportional hazard model. For the IPTW population, an adjusted KM estimate was used (RISCA package), and a robust variance estimator (survival package) was used to calculate the 95% confidence interval of the HR.¹⁴⁾ In sensitivity analysis, PS matching was utilized to control for differences in baseline characteristics. One-to-one matching was performed with the nearest neighbor (caliper: 0.2 of the SD of the logit of the PS) without replacement. A robust variance estimator (survival package) to calculate 95% confidence intervals for HRs was used in the PS-matched population.¹⁴⁾ All analyses were conducted using R version 4.2.1, and a 95% confidence interval for HR that did not cross 1 was considered statistically significant.

Ethical Considerations

This study was approved by the Ethics Committee in The University of Tokyo (Ref. No.2023135NIe). Informed consent was waived due to the anonymous nature of the data.

RESULTS

We identified 10189 CD-diagnosed patients in the JMDC database; including 4410 patients who received anti-TNFα treatment. Of these, 556 patients received vedolizumab or ustekinumab. However, 3 patients did not have a recorded treatment date for vedolizumab or ustekinumab, 215 patients began treatment with vedolizumab or ustekinumab before May 2019, 8 patients initiated anti-TNFα treatment before their CD diagnosis, and 31 patients started vedolizumab or ustekinumab before receiving anti-TNFα treatment: 50 in the vedolizumab group and 259 in the ustekinumab group. Among the vedolizumab group, 14 patients were excluded: 5 for the look-back period of fewer than 365 d, 6 for being under 18 years of age, 1 for pouchitis, and 2 for having the index date within the last observation month. Among ustekinumab group patients, 39 patients were excluded: 22 for the look-back period of fewer than 365 d, 9 for being under 18 years of age, 2 for pouchitis, and 6 for having the index date within the last observation month (Fig. 2).

Fig. 2. Flow Diagram of the Patient Selection Process

TNFα, Tumor necrosis factor-α; CD, Crohn’s disease.

Patient characteristics before and after IPTW adjustment are presented in Table 1. While prior exposure to infliximab and/or adalimumab was more prevalent in the ustekinumab group, and perianal disease and extraintestinal manifestations were more commonly observed in the vedolizumab group, following IPTW adjustment, these covariates were well-balanced. However, imbalances in sex and duration of CD persisted.

Table 1. Baseline Characteristics of the 256 Patients Included in the Main Analysis

n	Before IPTW			After IPTW
	Vedolizumab	Ustekinumab	SMD	Vedolizumab	Ustekinumab	SMD
	36	220		239.3	255.2
Male sex, n (%)	32 (88.9)	159 (72.3)	0.43	206.1 (86.1)	190.3 (74.6)	0.294
Age (years), mean (S.D.)	39.67 (14.04)	39.04 (11.55)	0.049	40.22 (15.26)	39.17 (11.70)	0.078
Anti-TNFα administration
Infliximab*, n (%)	14 (38.9)	144 (65.5)	0.552	144.3 (60.3)	158.0 (61.9)	0.033
Adalimumab*, n (%)	26 (72.2)	122 (55.5)	0.354	154.9 (64.7)	147.1 (57.6)	0.146
Infliximab and adalimumab, n (%)	4 (11.1)	46 (20.9)	0.27	60.0 (25.1)	49.9 (19.6)	0.133
Immunosuppressants, n (%)	10 (27.8)	51 (23.2)	0.106	48.2 (20.2)	60.5 (23.7)	0.086
Steroids, n (%)	12 (33.3)	62 (28.2)	0.112	73.3 (30.6)	75.1 (29.4)	0.026
CD-related surgery, n (%)	3 (8.3)	22 (10.0)	0.058	28.3 (11.8)	25.4 (9.9)	0.06
Perianal disease, n (%)	13 (36.1)	54 (24.5)	0.254	45.9 (19.2)	65.7 (25.8)	0.158
Extraintestinal manifestations, n (%)	17 (47.2)	73 (33.2)	0.289	93.7 (39.2)	89.6 (35.1)	0.084
Psoriasis, n (%)	2 (5.6)	12 (5.5)	0.004	6.3 (2.6)	13.7 (5.3)	0.139
CD diagnosis age 41years over, n (%)	8 (22.2)	34 (15.5)	0.174	55.7 (23.3)	42.7 (16.7)	0.164
CD duration (d), mean (S.D.)	3638.58 (2684.76)	3362.28 (2474.86)	0.107	2872.88 (2362.51)	3367.54 (2508.97)	0.203

S.D., Standard deviation; TNFα, Tumor necrosis factor-α; CD, Crohn’s disease; IPTW, Inverse probability of treatment weighting; SMD, Standardized mean differences. *Including patients who received both infliximab and adalimumab.

The treatment persistence for ustekinumab was significantly longer than the vedolizumab group in both the unadjusted population (HR, 0.33; 95% confidence interval (CI), 0.18–0.60) and IPTW adjusted population (HR, 0.32; 95% CI, 0.15–0.72) (Figs. 3A, B). A sensitivity analysis was conducted on the PS-matched population and also statistically significant difference was observed (HR, 0.25; 95% CI, 0.076–0.83) (Fig. 3C). To compare the steroid-free period, patient demographics were adjusted by IPTW following the exclusion of individuals using steroids at baseline. No statistically significant difference was observed between the two groups (HR, 0.38; 95% CI, 0.10–1.48, Fig. 3D), but the KM curve of ustekinumab was above that of vedolizumab. To evaluate the time to hospitalization, patient demographics were adjusted by IPTW after excluding patients who were hospitalized at the index date. No statistically significant difference was observed between the two groups (HR, 1.07; 95% CI, 0.60–1.91, Fig. 3E) and KM curves were largely overlapping. To evaluate the time to CD-related surgery, patient demographics were adjusted by IPTW following the exclusion of patients who underwent CD-related surgery in the look-back period. A statistically significant difference was observed between the two groups (HR, 0.33; 95% CI, 0.11–0.97, Fig. 3F). Patient characteristics, adjusted via PS matching for comparison of treatment persistence and IPTW for comparison of the steroid-free period, time to hospitalization, and time to CD-related surgery, are presented in Supplementary Materials 2–5.

Fig. 3. Results of Survival Analyses for A Treatment Persistence (Unadjusted), B Treatment Persistence Adjusted by Inverse Probability of Treatment Weighting (IPTW), C Treatment Persistence Adjusted by Propensity Score (PS) Matching, D Steroid Free Period Adjusted by IPTW, E Time to Hospitalization Adjusted by IPTW, F Time to CD-Related Surgery Adjusted by IPTW

HR, hazard ratio; CI, confidence interval; CD, Crohn’s disease.

DISCUSSION

This study compared the effectiveness of vedolizumab and ustekinumab in patients who failed to anti-TNFα treatment in Japan. Ustekinumab exhibited a significantly prolonged IPTW-adjusted treatment persistency, the primary endpoint. In the secondary endpoint and sensitivity analysis, the KM curves of ustekinumab in treatment persistency of the PS-matched population, steroid-free period, and time to CD-related surgery were above those of vedolizumab. In time to hospitalization, the KM curves of ustekinumab and vedolizumab largely overlapped. No outcome indicated that the KM curve of vedolizumab was above that of ustekinumab. These findings suggest that the effectiveness of ustekinumab was superior to vedolizumab in patients who failed anti-TNFα therapy even in Japan where the order in which these drugs were introduced in the market was different from that in the EU.

The KM curve for the steroid-free period in the vedolizumab group showed a rapid drop. This occurrence could be because of a highly weighted patient (high propensity score for ustekinumab treatment). Without this rapid drop, the KM curve would have largely overlapped. However, even if this were the case, the effectiveness of vedolizumab was not superior to that of ustekinumab. Rapid drops of the KM curve for treatment persistence and time to hospitalization adjusted by IPTW could also occur in the vedolizumab group. These rapid drops might also be attributed to a highly weighted patient. The results could be unstable due to the presence of highly weighted patient. Nevertheless, the overall trend remains unchanged, with the effectiveness of vedolizumab not surpassing that of ustekinumab.

Previous studies conducted in the EU demonstrated the longer treatment persistency of ustekinumab compared to vedolizumab,^5–8) as observed in this study. Therefore, these results may be considered generalizable regardless of the order of drug release. The proportion of steroid-free clinical remission was reported with inconsistent results in previous studies.^5–8) Some studies found that ustekinumab had a significantly higher steroid-free clinical remission proportion than vedolizumab and some studies found no statistical significance, but no study found vedolizumab to be superior. Our study measured the steroid-free period, and no statistically significant difference was observed, but the KM curve of ustekinumab was above that of vedolizumab, consistent with previous studies. A previous study did not find a significant difference in the proportion of hospitalization between vedolizumab and ustekinumab.⁵⁾ Our study also did not show a statistically significant difference in time to hospitalization, and the KM curves were largely overlapped. There is a possibility that the inclusion of non-CD-related hospitalizations obscured the difference, as we included all hospitalizations because the reason for hospitalization is not included in JMDC database. Alternatively, since we excluded patients who were already hospitalized at the index date, the sample size may have been too small to detect the difference. Our study found a statistically significant difference in the time to CD-related surgery between vedolizumab and ustekinumab. The previous study reported that there was no significant difference, but the point estimate of the odds ratio for abdominal CD-related surgery was below 1.⁵⁾ This can be attributed in part to the fact that nearly 50% of the patients had already undergone such surgery at baseline,⁵⁾ resulting in a low event proportion and wider confidence interval.

The order of drug launch in the U.S. is the same sequence as that in the EU—vedolizumab before ustekinumab.^15,16) A recent study in the U.S. examined the same clinical question.¹⁷⁾ Although no difference in treatment persistency was observed, ustekinumab outperformed vedolizumab in other outcomes. The reason for the lack of difference in treatment persistency remains unclear. However, with no outcomes demonstrating the superiority of vedolizumab, the results of this U.S. study are consistent with prior research from the EU and the current study.

Our study has several limitations. Firstly, it is a secondary analysis of JMDC claims database, which may be impacted by unmeasured confounding factors. The database lacks disease severity data, such as the Crohn’s disease activity index, Harvey–Bradshaw index, or CRP, which could potentially affect our results if there were differences in CD severity between the two treatment groups. However, we adjusted for various patient backgrounds, including past anti-TNFα administration, immunosuppressants, steroids, CD-related surgery, and perianal disease, which are considered proxies for CD severity. Thus, it is unlikely that there were significant differences in severity between the two groups after adjusting for these factors. Moreover, vedolizumab and ustekinumab were both recommended for patients who failed anti-TNFα treatment,⁴⁾ which may result in no substantial differences between the populations treated with the two drugs after anti-TNFα failure. Secondly, there is a possibility that unadjusted confounding factors may have influenced the results. Despite adjusting with IPTW, imbalances in gender and CD duration persisted in the main analysis. However, sensitivity analysis adjusted by PS matching successfully balanced these factors, and the results did not significantly differ from those of the main analysis. Therefore, we believe that the impact of remaining imbalances in gender and CD duration is not substantial. Thirdly, age at CD diagnosis and CD duration are inaccurate if the patient changed their health insurance society after the diagnosis of CD, as JMDC database lacks data before the change of health insurance societies. Fourthly, the target population of this study might have included patients who started vedolizumab or ustekinumab long after discontinuation of anti-TNFα treatment, who may have discontinued anti-TNFα treatment due to remission, rather than treatment failure. Because the reason for discontinuing anti-TNFα treatment is unclear, as JMDC database lacks such information. However, patients with a substantial gap between the discontinuation of anti-TNFα treatment and the initiation of vedolizumab or ustekinumab may be those who failed anti-TNFα treatment before the approval of vedolizumab or ustekinumab. Additionally, it is recommended to continue the same treatment for maintenance in patients with CD who achieved remission with anti-TNFα agents.⁴⁾ Therefore, we believe that the majority of patients in this study who discontinued anti-TNFα antibodies did so either due to ineffectiveness or adverse events. Finally, JMDC database does not include data on individuals over the age of 75, which may limit the generalizability of our study.

In conclusion, this study suggests that ustekinumab is a more effective treatment option than vedolizumab for patients who failed to anti-TNFα treatment. Notably, this finding remains consistent across both Japan and the EU. However, conclusive results cannot be established without conducting a head-to-head randomized controlled trial.

Acknowledgments

We would like to thank Dr. Yasuhiro Hagiwara for statistical advice.

Funding

This research was supported by Japan Agency for Medical Research and Development (AMED) under Grant Number: JP21lk0201701.

Conflict of Interest

MS is a full-time employee of Mitsubishi Tanabe Pharma Corporation; YM and DK declare no competing interests.

Data Availability

The JMDC claims database is available for anyone who purchases it from JMDC Inc. (https://www.jmdc.co.jp/en/jmdc-claims-database/).

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1) Roda G, Chien Ng S, Kotze PG, Argollo M, Panaccione R, Spinelli A, Kaser A, Peyrin-Biroulet L, Danese S. Crohn’s disease. Nat. Rev. Dis. Primers, 6, 22 (2020).
• 2) Roda G, Jharap B, Neeraj N, Colombel JF. Loss of response to anti-TNFs: definition, epidemiology, and management. Clin. Transl. Gastroenterol., 7, e135 (2016).
• 3) Masui S, Yonezawa A, Momo K, Nakagawa S, Itohara K, Imai S, Nakagawa T, Matsubara K. Infliximab treatment persistence among Japanese patients with chronic inflammatory diseases: a retrospective Japanese claims data study. Biol. Pharm. Bull., 45, 323–332 (2022).
• 4) Torres J, Bonovas S, Doherty G, et al. ECCO guidelines on therapeutics in Crohn’s disease: medical treatment. J. Crohns Colitis, 14, 4–22 (2020).
• 5) Alric H, Amiot A, Kirchgesner J, Tréton X, Allez M, Bouhnik Y, Beaugerie L, Carbonnel F, Meyer A. The effectiveness of either ustekinumab or vedolizumab in 239 patients with Crohn’s disease refractory to anti-tumour necrosis factor. Aliment. Pharmacol. Ther., 51, 948–957 (2020).
• 6) Biemans VBC, van der Woude CJ, Dijkstra G, van der Meulen-de Jong AE, Löwenberg M, de Boer NK, Oldenburg B, Srivastava N, Jansen JM, Bodelier AGL, West RL, de Vries AC, Haans JJL, de Jong D, Hoentjen F, Pierik MJ. Ustekinumab is associated with superior effectiveness outcomes compared to vedolizumab in Crohn’s disease patients with prior failure to anti-TNF treatment. Aliment. Pharmacol. Ther., 52, 123–134 (2020).
• 7) Townsend T, Razanskaite V, Dodd S, Storey D, Michail S, Morgan J, Davies M, Penman D, Watters C, Swaminathan M, Sabine J, Chapman A, Smith PJ, Flanagan PK, Reilly I, Bodger K, Subramanian S. Comparative effectiveness of ustekinumab or vedolizumab after one year in 130 patients with anti-TNF-refractory Crohn’s disease. Aliment. Pharmacol. Ther., 52, 1341–1352 (2020).
• 8) Manlay L, Boschetti G, Pereira B, Flourié B, Dapoigny M, Reymond M, Sollelis E, Gay C, Boube M, Buisson A, Nancey S. Comparison of short- and long-term effectiveness between ustekinumab and vedolizumab in patients with Crohn’s disease refractory to anti-tumour necrosis factor therapy. Aliment. Pharmacol. Ther., 53, 1289–1299 (2021).
• 9) Lenti MV, Dolby V, Clark T, et al. A propensity score-matched, real-world comparison of ustekinumab vs. vedolizumab as a second-line treatment for Crohn’s disease. The Cross Pennine study II. Aliment. Pharmacol. Ther., 55, 856–866 (2022).
• 10) Gisbert JP, Chaparro M. Ustekinumab to treat Crohn’s disease. Gastroenterol. Hepatol., 40, 688–698 (2017).
• 11) Pharmaceuticals and Medical Devices Agency (JP). “New Drugs Approved in FY 2016.”: ‹https://www.pmda.go.jp/files/000232770.pdf›, accessed 2 December, 2023.
• 12) Pharmaceuticals and Medical Devices Agency (JP). “New Drugs Approved in FY 2019.”: ‹https://www.pmda.go.jp/files/000235289.pdf›, accessed 2 December, 2023.
• 13) Nagai K, Tanaka T, Kodaira N, Kimura S, Takahashi Y, Nakayama T. Data resource profile: JMDC claims database sourced from health insurance societies. J. Gen. Fam. Med., 22, 118–127 (2021).
• 14) Austin PC. The use of propensity score methods with survival or time-to-event outcomes: reporting measures of effect similar to those used in randomized experiments. Stat. Med., 33, 1242–1258 (2014).
• 15) Ha C, Kornbluth A. Vedolizumab as a treatment for Crohn’s disease and ulcerative colitis. Gastroenterol. Hepatol. (N Y), 10, 793–800 (2014).
• 16) Shim HH, Chan PW, Chuah SW, Schwender BJ, Kong SC, Ling KL. A review of vedolizumab and ustekinumab for the treatment of inflammatory bowel diseases. JGH Open, 2, 223–234 (2018).
• 17) Kappelman MD, Adimadhyam S, Hou L, Wolfe AE, Smith S, Simon AL, Moyneur É, Reynolds JS, Toh S, Dobes A, Parlett LE, Haynes K, Selvan M, Ma Q, Nair V, Burris J, Dorand JE, Dawwas GK, Lewis JD, Long MD. Real-world evidence comparing vedolizumab and ustekinumab in antitumor necrosis factor-experienced patients with Crohn’s disease. Am. J. Gastroenterol., 118, 674–684 (2023).