Guide to the Application of Genotyping to Tuberculosis Prevention and Control
Combining Genotyping and Epidemiologic Data to Improve Our Understanding of Tuberculosis Transmission
Comparing Genotyping and Epidemiologic Results from Two TB Patients
In this section, we will analyze the most basic situation – the results of two TB patients. A conceptual framework for understanding the possible combinations of these results is provided in Table 4.5, which is separated into two categories: a) patients that were in fact involved in the same chain of recent transmission and b) patients that were not.
Genotypes Match and Known Epidemiologic Link Identified
If an investigation establishes that two persons with TB share a known epidemiologic link and their isolates have a matching PCR genotype, this provides good evidence that the two persons were involved in the same chain of recent transmission. This often means that one of the persons transmitted TB to the other, but it is also possible that both became infected by a third person. Other more complex transmission scenarios are also possible. Since the evidence strongly supports recent transmission, no additional information needs to be collected from a cluster investigation, and an RFLP analysis is not indicated.
Genotypes Match but No Epidemiologic Link or Only Possible Epidemiologic Link Identified
In contrast to the previous example, whether two persons are involved in the same chain of recent transmission is less clear if they have isolates with matching genotypes but routine case and contact investigations failed to identify a known link between them. If the two persons were, in fact, involved in the same chain of recent transmission, failure to identify when and where transmission occurred could be the result of investigators not asking the right questions or the persons interviewed not knowing or being unwilling to give complete answers. For example, if an outbreak occurs among drug users, it might take a lengthy investigation conducted by highly skilled interviewers to discover even a few of the actual epidemiologic links.
Another example of routine investigations being unable to discover hidden epidemiologic links are instances of casual contact. Casual contact was once thought to be a rare cause of TB transmission, but recent epidemiologic studies relying on genotyping results have shown that this concept needs to be updated. In a study conducted by the Maryland TB program, intensive interviews of 114 patients with matching genotypes and known source patients identified five patients who had only casual contact with the source patient (Cronin 2002).
The following are several possible scenarios in which two persons who have isolates with matching genotypes might not be involved in the same chain of recent transmission.
- The genotyping techniques used lack the power to discriminate between two isolates that are different. For example, if two isolates have matching PCR genotypes but different RFLP genotypes, the two isolates would be considered genetically different. If only the two PCR tests were conducted, however, and not the RFLP, the two isolates would be considered matched, even though they were genetically different.
- The transmission of common endemic strains of M. tuberculosis occurs in relatively closed populations. In a study conducted by the Arkansas Department of Health and CDC, experienced field workers interviewed 78 patients whose isolates had genotypes that matched those of at least one other person; epidemiologic links could be identified for only 33 (42%). The authors concluded that in a rural setting the presence of matching genotypes often is not associated with recent transmission (Braden 1997).
- An extensive outbreak of TB in the past leads to a large number of persons becoming latently infected with an identical strain of M. tuberculosis. Several years later these persons may reactivate their infections and develop active TB with matching genotypes. A contact investigation of these persons might well fail to establish an epidemiologic link between two patients, since the actual link connecting these patients occurred several years in the past. In a study by the New York City TB program of genotyping results for persons with MDR TB, of 153 persons who had matching genotypes, only 25 (19.8%) had epidemiologic links identified, and most persons had been exposed to patients diagnosed with TB before the study period, at a time when New York City was experiencing large outbreaks of MDR TB (Munsiff 2002).
- Cultures become contaminated in the laboratory and cause a false-positive culture to be reported. For example, during the processing of two specimens in a clinical laboratory, if a break in technique occurs and one specimen that contains M. tuberculosis is allowed to contaminate another specimen that does not, subsequent genotyping tests will report that the two specimens have isolates of M. tuberculosis that have matching genotypes.
Other types of laboratory error can occur and result in two isolates incorrectly being reported as having matching genotypes. For example, a submitting laboratory can mislabel specimens or ship the wrong specimen to the genotyping laboratory or the genotyping laboratory can mislabel specimens, perform the genotyping incorrectly, or mix up reports so that one specimen is reported to have the results of another specimen.
Genotypes Do Not Match but Known Epidemiologic Links Identified
As has been stated, there are multiple reasons why matching genotypes might not be associated with recent transmission. In contrast, it is much less common for isolates with nonmatching genotypes to be the result of recent transmission, since genotypes that do not match indicate that the two isolates are genetically different (unless there was a laboratory error in reporting those nonmatching results). Other than laboratory error, two possibilities have been reported where recent transmission was documented even though the isolates had nonmatching genotypes.
The first possibility occurs when the genotype of a particular strain of M. tuberculosis changes slightly over time. The DNA sequences on which genotypes are based have a tendency to mutate, change location, and be duplicated or deleted over time, forming the basis for the diversity of the genotypes. In another large outbreak of multidrug-resistant (MDR) TB in New York and surrounding areas, there was strong epidemiologic evidence that over 300 patients were infected with the same strain of M. tuberculosis (Bifani 1996). Genotyping with IS6110-based RFLP showed that the patients were infected with isolates that had multiple closely related genotypes. The investigators concluded that the M. tuberculosis strain that caused this outbreak had evolved during the prolonged outbreak. In practice, slight changes in RFLP patterns appear to be more common than changes with the PCR-based tests. Therefore, when a TB program requests that the genotyping laboratory run an RFLP on isolates in a PCR cluster, the genotyping laboratory report will flag isolates that have different but very similar RFLP patterns. Note that RFLP patterns will not be compared among isolates that do not belong to the same PCR cluster. Similarly, spoligotype patterns and MIRU results can change slightly over time. In this case, closely related isolates can show spoligotype or MIRU designations that differ by only one character. The genotyping laboratory report will not flag instances of PCR designations that differ slightly. However, if two patients have known epidemiologic links but have different PCR designations, the spoligotype and MIRU designations should be examined to determine if they are closely related. Isolates that match in all but one digit in the MIRU type or that have spoligotypes that differ in one region (as a result of loss of one or more spacers) should be considered closely related. Programs should consult with the genotyping laboratory or CDC to determine if RFLP typing of such isolates is warranted.
The second possibility occurs when a person is infected with more than one strain of M. tuberculosis. If a person with a dual infection transmits one strain to someone else, subsequent specimens from the two persons might grow the same strain of M. tuberculosis or might grow different strains, depending on which of the two strains grew from the specimen from the person with the dual infection. If different strains are grown from the two specimens, the isolates will have nonmatching genotypes. Although this situation has been reported, it is probably rare (Yeh 1999).
As mentioned, isolates with nonmatching genotypes are only rarely associated with recent transmission, even if epidemiologic links are identified. A much more common explanation of nonmatching genotypes where epidemiologic links have been established is when the links do not reflect recent transmission, and the presumed source case was not the real source of transmission. A study from the National Tuberculosis Genotyping and Surveillance Network documented common scenarios in which a presumed epidemiologic link between a source case and a secondary case was not confirmed by genotyping (Bennett 2002). Two common scenarios occurred when the presumed secondary case-patient was born in a foreign country or when the presumed source case-patient was culture-positive but sputum smear-negative.
Table 4.5. Conceptual framework for combining epidemiologic links with genotyping results for two tuberculosis cases. Percentages refer to the frequency that these categories were detected among 2,055 cases interviewed as part of the NTGSN study in the four sites that conducted cluster investigations of all TB patients with matching genotypes who did not have epidemiologic links identified during contact investigations (unpublished NTGSN data).
Nonmatching Genotypes with No or Only Possible Epidemiologic Links Identified
When isolates from two persons have nonmatching genotypes and interviews with the two persons reveal no known epidemiologic links, the two are probably not involved in the same chain of recent transmission. TB in both persons could represent reactivation of a latent TB infection, or one person might have been recently infected from an unidentified source case. This could happen when the source case did not live in the same TB jurisdiction or if no isolate was available for genotyping from the source case. No additional information needs to be gathered to strengthen this conclusion.