Of the 364 patients randomised (Fig. 1), 356 were included in the modified full analysis set (mFAS, comprising all randomised patients who met the inclusion criteria and received ≥ 1 dose of study medication). The treatment groups had similar baseline characteristics (Table 1) .
Rates of SCC at 30 days after end of treatment (EOT) (the primary endpoint) did not differ significantly between EPFX and standard vancomycin treatment in most of the subgroups analysed; however, in patients with C. difficile PCR-ribotype 027, rates were significantly higher with EPFX than vancomycin (Table 2).
SCC rates at days 40 and 90 were significantly higher with EPFX than with standard vancomycin in patients aged ≥ 75 years (Online Resource; ESM Fig. 1a). SCC rates at other time points and clinical response rates at day 12 and 2 days after EOT did not differ between treatments for either age category (Online Resource; ESM Fig. 1a and 1b). Our findings are supported by a previous study that used regression modelling to test the effects of age on treatment outcomes in phase 3 trials of standard-regimen fidaxomicin: a significantly higher probability of SCC was shown with fidaxomicin than standard vancomycin (odds ratio, 1.86; 95% CI, 1.40–2.47; P < 0.001). However, irrespective of treatment choice, the probability of SCC decreased by 13% and the risk of recurrence increased by 17% for each decade increase in age .
At day 40, rates of SCC were significantly higher with EPFX than with vancomycin in both the severe and non-severe CDI subgroups. At days 55 and 90, SCC rates were significantly higher with EPFX than vancomycin only in the non-severe CDI subgroup (Online Resource; ESM Fig. 2a). There were no significant between-treatment differences in efficacy at other time points in patients with severe and non-severe CDI (Online Resource; ESM Fig. 2b).
Patients with a previous history of CDI are at greater risk of developing a further CDI episode . Seventy-five (21%) patients in our analysis had experienced a recent CDI episode prior to study enrolment; the majority (55 [15%]) had a single prior episode. Rates of SCC at days 40, 55 and 90 were significantly higher with EPFX than with vancomycin in patients who had no prior occurrence of CDI. Additionally, rates of SCC at day 40 were significantly higher with EPFX in patients with one or two prior CDI episodes (Online Resource; ESM Fig. 3a). Other efficacy endpoints did not differ between treatments regardless of number of prior CDI episodes (Online Resource; ESM Fig. 3a and 3b). In a previous subgroup analysis, the 28-day SCC rate among patients with one prior CDI episode was 80.3% with standard fidaxomicin and 64.5% with standard vancomycin, higher than the rates observed in our analyses; this difference may be due to the younger age (median 63 years) of the patients in the previous analysis . Our conclusions are also limited by the low number of patients in our study with one or two prior CDI episodes, and the correspondingly high within-sample variability.
Patients without cancer at baseline had significantly higher rates of SCC at days 40, 55 and 90 with EPFX versus vancomycin (Online Resource; ESM Fig. 4a). Other efficacy endpoints in patients with and without cancer did not differ between treatments (Online Resource; ESM Fig. 4a and 4b). These results contrast with those from a previous analysis by Cornely et al. in which the SCC rate was significantly higher with standard-regimen fidaxomicin than vancomycin (73.6% versus 52.1%; P = 0.003) in 183 patients with CDI and cancer from two phase 3 studies . This difference in outcome may be attributable to the smaller sample size of our study (75 patients with CDI and baseline cancer) and slightly greater median age (75 years for the overall population in our study versus 69 and 63 years in patients with and without cancer, respectively, in the Cornely et al. study ). Our findings aligned with those of the previous study in that patients with cancer had lower rates of initial clinical cure and SCC than patients without cancer, regardless of treatment choice.
C. difficile PCR-ribotype 027 was the most prevalent ribotype in a recent pan-European survey  and has been associated with outbreaks of increased severity  and greater risk of CDI recurrence than other strains . In our study, rates of SCC at days 40, 55 and 90 were significantly higher with EPFX than with vancomycin in patients infected with C. difficile PCR-ribotype 027 (Fig. 2a). Rates of SCC at days 40 and 55 were significantly higher with EPFX than vancomycin in patients with other C. difficile PCR-ribotypes. No other significant between-treatment efficacy differences were observed in relation to C. difficile PCR-ribotype (Fig. 2a, b). In previous phase 3 registration studies, rates of recurrence were significantly lower in patients infected with non-PCR-ribotype 027 strains but not in patients infected with PCR-ribotype 027 [14, 15]. Further studies would be required to provide analyses of sufficient power to permit robust conclusions on efficacy differences in relation to PCR-ribotype 027.
Of note, clinical response at day 12 and 2 days after EOT was numerically lower with EPFX than with vancomycin in the majority of subgroups investigated here, and in the overall mFAS population . These results contrast with those of previous phase 3 trials [14, 15], in which standard-regimen fidaxomicin achieved numerically higher rates of clinical cure at 2 days after EOT, compared with standard vancomycin. The lower rate of initial clinical response with EPFX may be due to the administration of fidaxomicin on only alternate days after day 5, thus delaying the reduction in C. difficile count compared with vancomycin treatment.
Recurrence of CDI
Current ESCMID guidelines recommend that patients at risk of recurrent CDI are given standard-regimen fidaxomicin or vancomycin . Standard-regimen fidaxomicin is associated with a lower recurrence rate than standard vancomycin [14, 15]; moreover, in the overall EXTEND population, EPFX showed even lower recurrence rates than previously observed with the standard regimen . However, there was no between-treatment difference in the incidence of recurrent CDI according to the subgroups investigated here, although numbers were too small to permit definitive conclusions (Table 3).
The safety profiles of EPFX and standard vancomycin were broadly similar within the subgroups analysed. A greater number of treatment-emergent adverse events, serious adverse events and deaths were reported in patients aged ≥ 75 years in the vancomycin arm than in the EPFX arm (Table 4).