Comparison of multilocus variable number tandem repeat analysis and whole genome sequencing for investigation of Clostridium difficile transmission

No study to date has compared multilocus variable-number tandem-repeat analysis, MLVA, and whole-genome sequencing, WGS, for investigating Clostridium difficile transmission. Isolates from 61 adults with on-going/recurrent C. difficile infection and 17 asymptomatic carriage episodes in children (201 samples), and from 61 suspected outbreaks affecting 2-41 patients in 31 UK hospitals (300 samples) underwent 7-locus MLVA and WGS in parallel. Comparing first and last samples from the same individual taken median(IQR) 63(43-105) days apart, estimated rates of evolution of single nucleotide variants (SNVs), summed tandem repeat differences (STRD), locus variants (LVs) were 0.79(95%CI 0.00,1.75), 1.63(95%CI 0.00,3.59), 1.21(95%CI 0.00,2.67)/called-genome/year, respectively. Differences of >2SNVs and >10STRD have previously been used to exclude direct case-to-case transmission. Taking the first serial sample per individual to assess discriminatory power, across all pairs of samples sharing a PCR-ribotype, 192/283(68%) differed by >10STRD and 217/283(77%) by >2SNVs. Amongst all pairs of cases from the same suspected outbreak, using >2SNVs and >10STRDs to exclude transmission, 1190/1488(80%) pairs had concordant results. For discordant pairs, 229(15%) had =2SNVs but =10STRD, and 69(5%) had =2SNVs but =10STRD. Discordant pairs had higher numbers of LVs than concordant pairs, supporting the more diverse measure in each type of discordant pair. Conclusions on whether potential outbreaks were confirmed were concordant in 58/61(95%) of investigations. Overall findings using MLVA and WGS were very similar despite analysing different parts of the bacterial genome. With improvements in WGS technology it is likely MLVA loci data will be available from WGS in the near future.