In our study examining the diagnostic accuracy of the TST for detecting LTBI in a Canadian dialysis population, we found that the TST had poor sensitivity at both 5- and 10-mm thresholds of induration. Only 10 of the 62 patients who had a positive TST (and 5 patients with a negative TST) were found to have clinically diagnosed LTBI in the Manitoba TB and LTBI registries. Furthermore, only 8 of the 15 patients with LTBI were prescribed prophylaxis. In addition, one of three active TB cases had both a normal chest X-ray and negative TST result.
We hypothesize that the diminished sensitivity of the TST is likely a result of high rates of cutaneous anergy (40–50 %) reported in those with kidney failure [9, 11, 21]. Furthermore, false-positive results often occur in patients exposed to non-tuberculosis mycobacteria and the BCG vaccine [7, 11, 16, 17], also in concordance with our findings.
There are four drugs commonly used to treat tuberculosis: INH, RMP, pyrazinamide (PZA), and ethambutol (EMB), with INH being the first-line prophylaxis. The use of INH is not benign; there are many possible adverse effects including hepatitis, skin rash, flu-like syndrome, thrombocytopenia, and gastrointestinal upset. The incidence of adverse events from INH increases with age and is greatest in patients over 65 with comorbid conditions [22]. This may explain the small number of patients who were offered LTBI prophylaxis, as many patients may have been ineligible due to advanced age and comorbidities such as liver disease. Moreover, standard of care requires transplant candidates, who are often younger [23, 24], to receive the TST and prophylaxis if needed. This may have further contributed to the selection of younger patients offered pharmacologic therapy. Finally, these considerations highlight the importance of selectively screening only those patients who will actually qualify for prophylaxis, as recommended by the Canadian Tuberculosis Standards [22].
Our results are in keeping with previous findings from developing countries and highlight the low sensitivity and specificity of the TST in diagnosing LTBI in those with kidney failure [11, 16, 21], as well as the low rates of corresponding prophylaxis in those who do test positive with the TST [8, 25, 26]. A significant association was found between TST results and race, with white participants being less likely to test positive for LTBI than those from other ethnicities. This is not surprising because while overall TB rates in Canada have consistently declined, TB rates for Canadian-born aboriginal and foreign-born individuals are rising [22]. A similar trend has been noted in the USA with both overall and foreign-born TB rates declining but foreign-born rates decreasing at a much slower rate. Hispanic and Asian US immigrants may be affected up to 26 times more than non-Hispanic white individuals [27]. This phenomenon is likely due to factors such as immigration from endemic TB regions, generally lower socioeconomic status and educational resources, and more crowded living conditions among non-whites [28, 29].
While the TST is currently the most widely used diagnostic tool for LTBI, our results support the investigation of alternative diagnostic strategies for LTBI in dialysis patients. Several bodies now recommend using interferon-gamma release assays (IGRAs) as an alternative to TST in order to screen at-risk populations in high- and middle-income countries, including patients on dialysis and those awaiting transplant [22, 30]. IGRAs have increased sensitivity and specificity in dialysis patients; however, they are not typically employed in kidney failure patients [31], presumably because of the increased cost on a per-test basis [32].
IGRAs have many advantages over the TST as a diagnostic tool for LTBI. Like the TST, IGRAs assess the immune response to TB antigens, but with increased specificity [33], and although IGRAs are also subject to anergy [34], they may be less susceptible to uremic immunosuppression [33, 35]. Furthermore, the “boosting” effect that can been seen in patients subjected to repeated TST testing is absent [33, 35]; they require less technical acumen, eliminating diagnostic variability, and they are less cumbersome for the patient since a follow-up assessment is not needed. While the immediate costs of IGRAs are higher than those of TST, they may be more cost-effective in this population when considering long-term outcomes [32]. Some studies have shown that the sensitivity of IGRAs may be reduced post hemodialysis, but even under these conditions, IGRAs have a higher sensitivity than the TST [35, 36].
Our study has important clinical and research implications. First, since the TST has poor diagnostic performance for diagnosis of LTBI in those with kidney failure, it must be cautiously interpreted as a screening tool in dialysis patients. Alternatives such as IGRAs may perform better, but this needs to be confirmed in future studies in dialysis patients. Finally, screening for LTBI should only be performed in patients for whom the risk-to-benefit ratio of prophylaxis is favorable. Generally speaking, this would include younger patients and those waitlisted for kidney transplant, excluding the most elderly and highly comorbid patients for whom the risk of INH therapy may exceed the benefit of eradicating LTBI.
Our study has several strengths. We were able to provide a sample size of 483 patients who received a two-step TST, to date, one of the largest studies in a dialysis population from a non-endemic location [12, 37]. Through administrative database linkages, we were able to capture data informing which patients subsequently received pharmacological intervention. These data linkages enabled us to assess the clinical utility of the TST over a 5-year follow-up period.
There are also important limitations to our analysis. First, due to the lack of a true gold standard for diagnosis of LTBI [12, 19], we used radiographic evidence of previous TB infection or radiographic evidence plus risk of exposure and history of infection as the reference standard in our primary and secondary analyses, respectively. Although these criteria are known to be flawed reference standards, they have been widely applied in the published literature [12, 37], thereby facilitating comparisons with other studies and increasing the generalizability of our findings. Additionally, many of the risk assessment variables captured in our survey were self-reported and subject to patient recall bias. Many of these risk variables were also incompletely reported, as was the case for BCG vaccination status. Finally, due to the low number of patients who reported prior contact or history of TB infection, and their self-reported nature, we could not meaningfully analyze the sensitivity and specificity of these variables in isolation.