Characteristics of Tuberculosis Tests Performed during Postimport Quarantine of Nonhuman Primates, United States, 2021 to 2024

Inclusion criteria.

Shipment-level data were included for all legal NHP shipments with a final destination in the United States during FY 2021 to 2024. Shipments that only transited through the United States and NHPs that were confiscated while being smuggled into the United States were excluded. Individual animals were included if they were part of an eligible shipment and were euthanized during quarantine with clinical signs, ante/postmortem testing results, or exposure history suggestive of possible tuberculosis. Animals that tested positive only for nontuberculous mycobacteria were excluded from the analysis. We reviewed diagnostic test results submitted voluntarily by state health departments for imported NHP that were diagnosed with tuberculosis after quarantine, but these test results were not included in the analysis.

Antemortem testing.

Under 42 CFR 71.53, all imported NHPs are required to complete at least 3 negative TSTs, using MOT (Zoetis Animal Health, Parsippany, NJ), that are administered at least 2 wk apart, before they can be released from quarantine. All animals included in the analysis had completed at least one TST administered by the importer after arrival in the United States. We considered the animal’s most recent test before euthanasia or death, and results were recorded as positive (grade 3 or higher) or negative (grade 2 or lower).⁹ Some animals were tested with a proprietary IGRA test developed and performed by a single US commercial laboratory. Other animals underwent serologic testing with a multiplexed fluorometric immunoassay (MFIA) that targets 5 recombinant antigens; all samples were tested using the same commercially available test. The TST is a regulatory requirement, so all TST results were reported. IGRA and MFIA are elective tests, so reporting of positive results was mandatory, but reporting of negative results was voluntary.

Postmortem testing.

All NHPs included in the dataset underwent necropsy. In accordance with regulations, postmortem evaluation included histopathology of the lung, tracheobronchial lymph node, liver, and spleen, as well as other grossly abnormal tissues. If abnormalities consistent with tuberculosis were noted on histopathology, acid-fast staining was performed, and fresh tissues were submitted to the NVSL for mycobacterial culture to make a final determination on the animal’s tuberculosis status. Because histopathology and culture are required by regulation, all results (positive and negative) were reported. Samples submitted to NVSL for mycobacterial culture typically also undergo direct real-time PCR testing for Mycobacterium tuberculosis complex (MTBC) using IS1081.¹⁰ If samples are positive on MTBC culture, isolates also undergo whole genome sequencing (WGS).^d While reporting of these tests is not mandatory unless positive, results were typically available because they were included in culture reports.

NHPs were considered confirmed positive only if they had a positive MTBC culture. Because the NVSL MTBC PCR is neither validated for NHPs nor an approved test for regulatory purposes, animals with a negative culture and positive PCR were classified as having suspected TB in this analysis. Animals with a negative culture and positive staining for acid-fast bacilli (AFB) were also classified as having suspect TB because of the possibility of infection with other, non-MTBC acid-fast organisms.

Analysis.

We calculated sensitivity, specificity, and positive predictive value for TST, IGRA, MFIA, and MTBC PCR, using MTBC culture results as the gold standard. If an IGRA or MFIA test was indeterminate, the individual result was excluded from the calculation of that test’s performance characteristics. Animals that were positive for non-MTBC mycobacteria were excluded. Test characteristics are not reported for MFIA because of the small sample size. Because the calculation of negative predictive value relies heavily on negative animals, which are underrepresented in this sample, negative predictive value is not reported for any test. The CI for test characteristics was calculated in R Statistical Software (v4.1.2; R Core Team, 2021) using the DescTools package.^e To assess intertest agreement, we used the Cohen κ coefficients, which were calculated using the irr package.^f Comparisons were considered statistically significant at P ≤ 0.05.

NHP shipment characteristics.

Overall, 4% of shipments received had at least one TB-infected NHP identified during FY 2021 to 2024 (Table 1); however, the proportion varied by year with 1% to 4% of shipments impacted during FY 2021 to 2023 and 9% in 2024. During the analysis period, 5 (20%) of 25 registered importers received at least one shipment containing a TB-infected NHP. All affected shipments originated from 3 (8%) of 40 suppliers, all located in Southeast Asia. The median size (360 animals) of shipments with a NHP having confirmed tuberculosis was much larger than the median size (120 animals) of shipments without confirmed tuberculosis, and all affected shipments consisted of cynomolgus macaques (Macaca fascicularis) imported for scientific purposes.^g While the majority (10 out of 17, 58%) of shipments with confirmed tuberculosis had more than one animal affected, the overall proportion of infected animals in affected shipments remained low (median: 0.6%, range: 0.1% to 5.0%). In accordance with CDC regulations, the quarantine period was extended for all affected shipments, which lasted a median of 111 d (range: 84 to 314), compared with 39 d (range: 39 to 134) for unaffected shipments. With the exception of one shipment that had 2 separate animals infected with Mycobacterium caprae and Mycobacterium orygis, all NHPs with TB in affected shipments were infected by a single species of mycobacteria, and WGS indicated that all infected animals in each affected shipment had closely related isolates (≤5 single-nucleotide polymorphism differences). Seven shipments had NHPs found to be infected with M. caprae, 5 with M. orygis, and 6 with M. tuberculosis, all of which are part of the MTBC.

Diagnostic evaluation of NHP.

Every animal in the dataset (n = 153) received at least one tuberculin skin test (Table 2). Other antemortem tests not required by the regulation (that is, IGRA and MFIA) were each performed in less than 30% of animals. Staining for AFB was the most common postmortem test, with 145 animals (95%) receiving acid-fast staining of at least one tissue. Seventy-eight percent of animals also had MTBC PCR and MTBC culture performed on at least one tissue.

A total of 69 NHPs (0.07% of all NHPs imported during this period and 1% of NHPs from shipments with confirmed TB) had a positive MTBC culture, and an additional 15 animals (not shown) were suspected to have tuberculosis based on positive acid-fast staining, MTBC PCR, or both. Of the 69 animals with positive cultures, WGS identified 38 (55%) that were infected with M. orygis, 20 (29%) with M. caprae, and 11 (16%) with M. tuberculosis. Three animals were infected with non-MTBC mycobacteria, and these animals were omitted from the analysis; all 3 were positive on TST and negative on AFB staining and MTBC PCR of tissues.

Diagnostic test characteristics.

Overall, TST had the highest sensitivity and lowest specificity and positive predictive value of all tests (Table 3). MTBC PCR had the highest specificity and positive predictive value (with a prevalence among tested animals of 44%) of all tests.

Among the antemortem tests, TST and IGRA were both more sensitive than specific. TST had a sensitivity of 94% (95% CI: 86% to 98%), specificity of 18% (9% to 31%), and positive predictive value of 61% (51% to 71%). IGRA had a sensitivity of 67% (46% to 83%), specificity of 45% (17% to 77%), and positive predictive value of 75% (53% to 90%).

Among the postmortem tests, acid-fast staining was more sensitive than specific, with a sensitivity of 81% (70% to 89%), specificity of 69% (54% to 81%), and positive predictive value of 79% (67% to 87%). MTBC PCR had a sensitivity of 84% (73% to 92%), specificity of 82% (69% to 91%), and the highest positive predictive value of any test (87%, 76% to 94%). A summary of positivity rates by tissue type is included in Table S1.

Intertest agreement.

Kappa scores were statistically significant for 5 comparisons: TST compared with MFIA (κ = −0.29, no agreement), TST compared with acid-fast staining (κ = 0.22, minimal agreement), acid-fast staining compared with MTBC PCR (κ = 0.63, moderate agreement), acid-fast staining compared with culture (κ = 0.54, weak agreement), and MTBC PCR compared with culture (κ = 0.67, moderate agreement; Table 4). Of these, only the comparisons between postmortem tests represented better than minimal agreement based on standard interpretation of κ statistics.¹¹ Kappa was also calculated for TST compared with IGRA including results for 269 animals from a single shipment that all received IGRA testing; most were not euthanized or necropsied and therefore were not eligible for inclusion in the rest of the analysis. All 269 animals tested negative on TST and IGRA, and including them improved the TST compared with IGRA κ from −0.15 to 0.69 (moderate agreement; P < 0.001).

Postquarantine cases.

The CDC is aware of 16 NHPs from 7 shipments that were diagnosed with tuberculosis after completing CDC-mandated quarantine during this time period. These cases were identified a median of 14 mo (range: 1 to 27) postquarantine. Eight (50%) cases occurred in animals from 2 shipments that had TST-positive animals identified during quarantine. Mycobacterial species information was available for 10 postquarantine cases: 6 were infected with M. tuberculosis, 2 with M. orygis, and 2 with M. bovis. For the 7 cases with available data, the species of mycobacterium detected postquarantine was the same as the species detected in NHPs from the same shipment that tested positive during quarantine.

Whole genome sequencing was available only for the 4 animals infected with M. bovis or M. orygis. The animals with M. bovis came from the same supplier in Africa, but in different shipments, and neither shipment had animals with TB identified during quarantine. The isolates were genetically identical to each other. The 2 animals infected with M. orygis originated from the same supplier in Southeast Asia but were from different shipments and were detected at different postquarantine facilities. The 2 postquarantine M. orygis isolates were identical to isolates previously detected during quarantine in other animals from the same supplier. One of the animals came from a shipment with TB identified during quarantine, but the isolates detected during quarantine differed from the postquarantine isolate by 8 single-nucleotide polymorphisms.

Limitations.

The use of observational (compared with experimental) data to conduct this analysis allows the data to reflect real-world conditions, but it also comes with significant limitations. Many diagnostic tests perform differently depending on how recently an animal was infected, but the observational nature of this investigation did not allow us to determine when an animal was infected, so this could not be considered during the analysis. It is also likely that the different TB diagnostics were conducted in series for some animals, and the results of previous tests may have affected the likelihood that specific animals or tissues would receive subsequent tests. While this analysis reflects the typical diagnostic process for these animals, it would be preferable to calculate test characteristics from a dataset that was not biased in this way. The bias this produces is especially apparent for animals that tested negative on antemortem tests; these animals are much less likely to receive the gold standard test (MTBC culture), so they are underrepresented in the dataset, and all test characteristics that include them are less reliable as a result. Negative results on tests that are not required by regulation may also be underreported, since reporting of these results is voluntary. Calculation of specificity and negative predictive value depends on data about negative animals, and this bias might partially explain why the specificity of the TST and IGRA were so different from previous reports.⁴ We used culture as the gold standard for calculations of test characteristics because it is used as the official confirmatory test in the regulations, but it is important to note that false negatives are possible. Positive predictive values and our assessment of intertest agreement represent a reasonable approximation for how well tests agree for animals with a higher degree of suspicion for TB but may not generalize well to settings with a different TB prevalence. It is also important to note that many infected animals in this analysis came from the same shipments, and thus their data are not independent. Which shipment an animal was part of likely affected what ancillary diagnostics it received, and the different sizes of outbreaks likely affected the proportions of different species of MTBC identified.

We included voluntary reports of imported NHPs testing positive after completing quarantine as evidence supporting the assertion that current TB screening protocols lack sufficient sensitivity given the change in TB epidemiology. However, the voluntary nature of TB reporting after quarantine precludes the use of these data to assess the true proportion of imported NHPs diagnosed with TB after quarantine.

Conclusions.

Given the change in TB epidemiology in imported NHPs, there is a need for more sensitive and specific antemortem TB screening tests to improve detection and minimize quarantine duration. For postmortem testing, MTBC PCR, while currently not approved for confirmation for regulatory purposes, has a relatively high positive predictive value and might represent a valuable tool to support decision-making while culture results are pending. Mandatory reporting of postquarantine cases of TB in imported NHPs to states, with subsequent notification to the CDC, would allow a more complete assessment of the potential public health impact of failure to detect TB in imported NHPs during quarantine.