Dear Editor,

Most TB in the United States arises from reactivation of past Mycobacterium tuberculosis infection.^1–4 Identifying and treating recently infected persons are priority prevention and control activities because recently acquired M. tuberculosis infection can rapidly progress to TB disease.⁵ Therefore, health departments in the United States routinely use local TB genotyping results to gauge their success in suppressing recent transmission (RT). According to PCR-based genotyping methods used for over a decade, approximately 12% of all TB appears attributable to RT.^4,6,7 In 2018, whole-genome sequencing (WGS) became routine for each culture-confirmed TB case, and in 2022, PCR-based genotyping (i.e., GENType designations) was discontinued. These new WGS methods produce a whole-genome multilocus sequence typing (wgMLSType) designation. wgMLSType is assigned by comparing the sequences of 2,672 genetic loci, chosen based on presence in ≥95% of genomes in a test dataset of 5,000 genomes, and assigning the same wgMLSType to cases that have matching patterns at ≥99.7% of the loci, a cutoff that ensures clustering of isolates that are within seven single nucleotide polymorphisms (SNPs). However, wgMLSType designation does not provide the number of pairwise SNP differences between isolates, and cases with a matching wgMLSType might not necessarily be closely related enough genetically to be consistent with RT. As recommended in a recent publication regarding best practices for WGS,⁸ whole-genome SNP comparison and phylogenetic trees, which are more discriminatory than wgMLSType, are produced by CDC for a select subset of wgMLSType clusters. These are used along with timing, geography, and other epidemiologic characteristics to further assess the likelihood that cases are related by RT for cluster investigation.

Using TB cases reported to the National Tuberculosis Surveillance System⁴ from 2018–2021 (i.e., the 4 years when both GENType and wgMLSType designations were assigned), we produced new sequencing-based RT estimates for the 2020–2021 cases. We applied the previously validated plausible source case method; that is, if ≥1 plausible source case with a matching wgMLSType designation was identified during the previous 2 years within 10 miles, the TB case was attributed to RT.^6,7,9 We compared the proportion of TB cases attributed to RT using the older GENType and newer wgMLSType designations, both in the overall population and stratified by origin of birth. We identified counties with ≥10 TB cases during 2020–2021 that had ≥10% absolute difference in RT using the two genotyping methods. We described those counties by their rural-urban classification (U.S. Department of Agriculture),¹⁰ poverty level (U.S. Census American Community Survey),¹¹ and annual TB count per 100,000 persons (National Tuberculosis Surveillance System).⁴

We found that with the PCR-based GENType designation, 12.3% of TB cases nationally were attributed to RT; the corresponding proportion was 11.8% using the sequencing-based wgMLSType designation. Among U.S.-born persons with TB, the proportion of TB cases attributed to RT increased from 24.8% to 28.2%. Conversely, RT estimates decreased from 7.7% to 5.9% among non-U.S.-born persons. In total, 14 U.S. counties reported ≥10 TB cases during 2020–2021 and had ≥10% absolute difference in the proportion of cases attributed to RT. The Table shows that estimates of RT increased ≥10% in 10 counties and decreased ≥10% in 4 counties. The 10 counties with new higher RT estimates varied widely in population size, composition, and annual TB counts, but nearly all had substantial poverty. In these 10 counties, 37 persons whose TB was previously not attributed to RT were reclassified as having TB attributed to RT because wgMLSType grouped closely related isolates with different PCR-based GENTypes. All 37 were either U.S.-born (n = 22) or Mexico-born (n = 15). The four counties with new lower RT estimates were all urban with larger populations and generally lower poverty levels. Among the 36 total persons reclassified as not having TB attributed to RT, 29 (80.6%) were born in China or the Philippines. However, small absolute numbers drove the ≥10% of case reclassifications in these 14 counties.

In conclusion, the overall proportion of TB cases attributed to RT was similar with either genotyping method, but the sequencing-based wgMLSType designation produced a somewhat higher RT estimate among U.S.-born persons and a somewhat lower estimate among non-U.S.-born persons. Poverty in counties with increased RT estimates suggests fewer resources and greater challenges for TB prevention and control at the local level. This poverty context also emphasizes the importance of the direct connections with the communities most affected by RT and including those voices in developing public health strategies to decrease RT.⁸ Transition to WGS-based genotyping will require changes in methods to estimate RT to ensure accuracy; accurate RT estimates will facilitate assessments of efforts to suppress RT. Further work is necessary to validate the findings of this analysis in the context of local epidemiology and to optimize RT estimates for WGS-based genotyping. In the meantime, these findings of similar overall RT estimates in the context of differences by birth origin and specific locations and settings might be helpful context for public health decision-making.