Collagenase Injection versus Limited Fasciectomy for Dupuytren's Contracture

Abstract

Background: Treatments for Dupuytren's contracture include limited fasciectomy and collagenase injection. Comparisons of the effectiveness of these treatments have been limited.

Methods: We performed an unblinded, multicenter, pragmatic, two-group, randomized, controlled noninferiority trial comparing collagenase injection with limited fasciectomy in persons with moderate Dupuytren's contracture. The primary outcome was the score on the Patient Evaluation Measure-Hand Health Profile (PEM), a questionnaire for assessing hand health as reported by the patient, at 1 year after treatment. Scores on the PEM range from 0 to 100, with higher scores indicating worse outcomes. The prespecified noninferiority margin was 6 points.

Results: A total of 672 persons (336 per group) were assigned to receive collagenase injection or to undergo limited fasciectomy. The primary analysis included 599 persons: 314 in the collagenase group and 285 in the limited-fasciectomy group. The mean score on the PEM at 1 year was 17.8 among the 284 patients with available data in the collagenase group and 11.9 among the 250 patients with available data in the limited-fasciectomy group (estimated difference, 5.9 points; 95% confidence interval [CI], 3.1 to 8.8; one-sided P = 0.49 for noninferiority). Among the patients with available data (229 patients in the collagenase group and 197 patients in the limited-fasciectomy group), the estimated difference in the mean score on the PEM at 2 years was 7.2 points (95% CI, 4.2 to 10.9). Moderate or severe complications of treatment occurred in 1.8% of the patients in the collagenase group and in 5.1% of those in the limited-fasciectomy group; recurrent contracture resulted in reintervention in 14.6% and 3.4%, respectively.

Conclusions: Collagenase injection was not noninferior to limited fasciectomy with respect to the score on the PEM at 1 year after treatment. (Funded by the National Institute for Health and Care Research Health Technology Assessment Programme; DISC ISRCTN Registry number ISRCTN18254597.).

Figure 1. Participant flow diagram.

Participants are categorized as censored if the trial follow-up period finished prior to the follow-up time point being due/completed

Figure 2. Estimated differences in expected scores and two-sided 95% confidence intervals over time for the four participant reported outcome measures.

The vertical grey dashed line is plotted at the median time elapsed between baseline and treatment delivery, with subsequent time points referenced to this. For plots (a) and (b), values above zero indicate greater benefit from LF. For plots (c) and (d), values above zero indicate greater benefit from collagenase. 95% CIs are not adjusted for multiplicity and should therefore not be used for hypothesis testing with respect to any of the secondary endpoints. Figure 2a: Patient Evaluation Measure. The estimated differences at each time point are: 3 months = -3.8 (95%CI -6.2 to -1.3), 6 months = 1.4 (-1.2 to 4.1), 1 year = 5.9 (3.1 to 8.8), 2 years = 7.5 (4.2 to 10.9). The non-inferiority margin for the primary endpoint (Patient Evaluation Measure at 1 year) was +6 points. Figure 2b: Unité Rhumatologique des Affections de la Main scores (numerical estimates provided in supplementary material) Figure 2c: Michigan Hand Questionnaire scores (numerical estimates provided in supplementary material) Figure 2d: Single Assessment Numeric Evaluation scores (numerical estimates provided in supplementary material)

Figure 3. Estimated differences in passive and active extension deficit.

The vertical grey dashed line is plotted at the median time elapsed between baseline and treatment delivery, with subsequent time points referenced to this. For all plots, values above zero indicate greater benefit from limited fasciectomy. Measurements were carried out by trained surgeons, therapists or research nurses using a goniometer and supported by a goniometry measurement manual. The interval estimates for these secondary endpoints have not been adjusted for multiplicity and should not be used for hypothesis testing.