Interferon-γ Release Assays for Diagnosing Mycobacterium tuberculosis Infection in Renal Dialysis Patients

Materials and Methods

This study was part of a multisite study conducted by the U.S. Centers for Disease Control and Prevention (CDC) to evaluate new tests for LTBI. The study protocol was approved by the institutional review boards of CDC, California Department of Health Services, and other participating study sites.

For this investigation, we toured the dialysis center and interviewed patients and staff. We reviewed the TB case-patients’ charts for clinical details and dialysis history. Dialysis logs were reviewed for patients’ dialysis weekly schedules, seating arrangements, and history of TST results. The amount of daily dialysis time shared between the TB case-patient and each other patient was calculated in 15-min intervals in 1- and 3-mo periods before he was removed from the dialysis center. Overlapping time periods of less than 15 min were not counted. The patient had no pulmonary symptoms until developing a cough during the last week before his exit from the center. During that week he dialyzed daily (Monday through Friday). Before this week, and for the 3 mo before his exit, he dialyzed three times weekly (Monday/Wednesday/Friday). Accordingly, and a priori, we decided that any patient who shared dialysis time with him during his final and symptomatic week was a “contact.” These turned out to be patients who had a shared Monday/Wednesday/Friday dialysis schedule in the month before his departure, as well as patients with Tuesday/Thursday/Saturday schedules who dialyzed with him on the Tuesday and Thursday of his last week. All other patients were considered “noncontacts.”

The center had a large open floor divided in half by a 4-ft median, with dialysis stations positioned on either side (Figure 1). We also attempted to assess the infectious risk of TB case-patient contact with a trichotomous exposure variable “proximity.” Noncontacts were considered “A”, contacts seated on the opposite side of the room as the case-patient were “B”, and contacts seated on the same side as the case-patient were “C.” There was one station on the case's side of the room (Figure 1, bottom, right station on diagram) that was located farther away than the closest station on the other half of the room. Accordingly, this station was considered “B” for the purposes of this analysis. This variable was analyzed as a categorical variable in logistic regression. Similar proximity analysis using A (no contact), B (seated ≥15 ft from case), or C (seated ≤15ft from case) produced similar results and is therefore not presented. Air flow in the facility was not assessed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Diagram of dialysis center. A 4-foot barrier divided the room in half (shaded vertical box). The small boxes arranged on either side represent dialysis “stations” or seats. The TB case-patient's station is in black.

Patients had laboratory work performed monthly as part of their routine renal care. We collected patients’ most recent predialysis blood urea nitrogen, hemoglobin concentration, and serum creatinine and albumin levels (all labs had been drawn 1 to 3 d before enrollment).

We became aware of the situation nearly 10 wks after the TB case-patient was removed from the center. At 11 wks, we arrived at the center and asked all patients there to participate. Patients were excluded from the study if they had a documented history of a positive TST result ≥10 mm or a history of prior treatment for LTBI or TB disease. After obtaining consent, we drew 30 ml of blood before the subject was connected to the hemodialysis machine. Blood was transported to the Long Beach Public Health Laboratory (Long Beach, CA) at room temperature where QFT-G was performed and interpreted as described previously (12). For the ELISPOT assay, blood was collected in a BD Vacutainer CPT cell preparation tube with sodium citrate and centrifuged for 20 min according to the manufacturer's instructions (BD Biosciencs, Franklin Lakes, NJ) and shipped overnight to the laboratory in Portland, OR where the peripheral blood mononuclear cells were prepared and cryopreserved for future analysis. ELISPOT was performed according to published methods (13,14). After venipuncture, patients underwent routine Mantoux TST with 0.1 ml (5 TU) of tuberculin purified protein derivative (Tubersol, Connaught Laboratories Limited, Toronto, Ontario, Canada). TSTs were read 48 to 72 h later by study personnel. Induration of ≥5mm was considered a positive TST result. Positive QFT-G assay results were defined per QuantiFERON-GOLD® test-kit manufacturer's instructions (15). Positive ELISPOT results were determined in the following fashion. First, to determine the frequency of antigen specific responses, 250,000 peripheral blood mononuclear cells were incubated with both ESAT-6 and CFP-10 antigens. These antigens were comprised of pooled peptides reflecting the entire gene. Each peptide pool was comprised of 15 amino acid peptides, and each peptide overlapped by an 11-amino acid segment with the adjacent peptide (5 μg/peptide per ml; Genemed Synthesis, South San Francisco, CA). Controls included media (nil) and phytohemagglutinin 10 μg/ml (Sigma-Aldrich, St. Louis, MO). All determinations were performed in duplicate. For the ELISPOT assay, spots were counted by an automated ELISPOT reader (Zeiss). Lab personnel were blinded to clinical and epidemiologic data. For a response to be considered evaluable, the mean number of spots (reactive lymphocytes) had to be 3 SD greater than the media control well. If evaluable, then the response of the media control well was subtracted from the antigen specific response. A result greater than 10 spots was considered positive.

To assess the reproducibility of the blood tests, we returned at 16 wk after exposure to draw more blood for QFT-G and ELISPOT. TST was also repeated at this time if the initial TST response was <5 mm. Consistent with the principles of TB contact investigation, patients were defined as “TST positive” if they had positive TST results on either initial or repeat assessment. Similarly, persons with a positive result on either the initial or repeat blood testing were considered “positive” by IFN-γ assay.

All data were entered into Epi-Info (version 3.4.1, CDC, Atlanta, GA). Univariate logistic regression was used to assess the association of various patient characteristics with the outcome of skin or blood test results. Multivariate models were constructed using variables that, in univariate analysis, were associated with a test result with a P ≤ 0.20. Because of significant differences in age between contacts and noncontacts, age was included in all multivariate models. Stepwise backward logistic regression was then performed using Fischer exact P values (P values ≤0.05 were considered statistically significant).

Results

The index patient was a 64-yr-old foreign-born man. In the month before his diagnosis, he routinely dialyzed on a Monday/Wednesday/Friday schedule in the first of three shifts (5 a.m. to 9 a.m.) offered on those days (there were 4 daily shifts on Tuesday and Thursday). Approximately 1 mo before TB diagnosis, he developed ascites. One week before his TB diagnosis, he developed a productive cough. Because of worsening ascites, he attended dialysis every day (Monday through Friday) in the last week before being removed from the center, admitted to the hospital, and diagnosed with TB disease. At that time, his sputum was smear-positive for acid-fast bacilli. His chest radiograph showed pleural effusion and parenchymal infiltrate. M. tuberculosis was isolated from sputum and from peritoneal fluid. The isolate was sensitive to all first-line anti-tuberculosis drugs tested.

We attempted to enroll all dialysis patients from the first three 4-h dialysis shifts (5 a.m. to 9 a.m., 9 a.m. to 1 p.m., 1 p.m. to 5 p.m.) of each day (n = 124), and 100 were enrolled. Twelve declined and 12 were excluded because of prior positive TST results or previous LTBI treatment. During our statistical analysis, 6 additional patients were excluded from QFT-G analysis because of improper blood specimen labeling (n = 4), or indeterminate results (n = 2). Three patients were excluded from ELISPOT analysis for having only indeterminate ELISPOT results. The total number of patients with evaluable results for each test were: TST (n = 100), QFT-G (n = 94), and ELISPOT (n = 97). The median age of patients was 55 yr (range, 18 to 90 yr), 47 (47%) were born outside the United States or Canada, and 55 (55%) were male. The most common countries of origin for foreign-born patients were: Mexico (n = 19), Philippines (n = 12), and Cambodia (n = 5). Seven were from American Samoa. Fifty-eight (58%) patients were considered contacts. In the month before his diagnosis, contacts had an average of 18 h of exposure (range, 1 to 52 h) to the case, whereas noncontacts had less than 0.5 h of exposure (range, 0 to 6 h) and no contact during the index case's symptomatic week. Contacts were similar to noncontacts with regard to sex and foreign birth, but they were significantly younger than noncontacts (Table 1).

Discussion

We compared two whole-blood based IFN-γ assays with tuberculin skin testing in ESRD patients during a TB contact investigation at a hemodialysis center. A high percentage of persons were found to be positive regardless of the test used. IFN-γ assay, but not TST, results was correlated with contact to the TB case-patient in the facility. Similar results from both of these assays suggested the possibility of TB transmission between the TB case-patient and contacts. Our data suggested that TST results were influenced by “boosting” and that TST might not be as useful as IFN-γ assays in evaluating hemodialysis patients with recent exposure to TB.

QFT-G and ELISPOT results were most closely associated with TB case-contact. Results were similar when we analyzed the risk of positive tests results according to proximity of seating between contacts and the case-patient. These associations between positive test results and TB case-patient contact were independent of other TB risk factors and suggested the possibility that TB transmission occurred in the dialysis facility. Conversely, analysis of TST results did not suggest TB transmission. A possible explanation for this finding is that the IFN-γ assays are more sensitive and/or specific for recent TB infection than the TST in this study population.

Previous studies have suggested high rates of anergy to TST among ESRD patients (8,9). This is the first study, to our knowledge, that has assessed the ability of IFN-γ assays to detect recent TB infection in renal dialysis patients. Without a gold standard for TB infection, it is impossible to determine which test is more accurate and less influenced by the immune suppression of ESRD. Interestingly, individuals in our study with a positive QFT-G assay result and negative TST result were significantly more likely to be TB case-patient contacts than those persons with concordant negative test results. One possible explanation is that the QFT-G assay might be more sensitive in detecting recent infection in this population. Such improved sensitivity could be the result of a number of factors, including the possibility that in vitro testing of IFN-γ response is less affected by uremic immune suppression or malnutrition than in vivo assessment with the TST. This has been shown previously with immune responses to purified protein derivative (PPD). Sester et al. (16) compared ESRD patient immune response to PPD using both in vivo (the TST) and in vitro methods (flow-cytometric quantification of PPD-specific reactive T cells). They found that a large percentage of persons who had reactive in vitro assays lacked positive TST results, suggesting a high rate of cutaneous anergy to the TST in their dialysis population (16). Furthermore, several studies in immunocompetent persons suggest that blood-based IFN-γ assays using certain RD-1 antigens are better correlated with recent TB exposure than TST, and therefore, possibly more sensitive in detecting recent TB infection (17,18).

The use of highly specific antigens for M. tuberculosis in the IFN-γ assays provides the theoretical basis for a more specific test for LTBI compared with the PPD-based TST. This has been suggested in a number of studies of immunocompetent subjects using the same QFT-G assay used in our study (11,19–21) and similar RD-1 antigens in an ELISPOT platform (16,22). Only one published study compared TST with either the QFT-G or T-SPOT (a commercialized ELISPOT assay) in ESRD patients. In that study, there was no defined TB exposure, so they were limited in their ability to assess which test result was more likely to represent TB infection (23). Although limited by low study numbers, our analysis of patients with discordant TST and QFT-G results supports the possibility that TST was less specific for detecting recent TB infection in the dialysis center. Persons with positive TST and negative QFT-G results were more likely (although not quite statistically significant) to have been foreign born than those with concordant negative TST and QFT-G results. In some of these persons, BCG vaccination might have caused false-positive TST results. BCG has been shown to strongly influence TST results up to 10 mm of induration in patients older than 40 yr (10). In previous studies comparing INF-γ assay and TST results, similar discordance has been associated with previous BCG vaccination or nontuberculous mycobacterium exposure (11). Although we attempted to collect BCG vaccination histories from the patients in this study, because of language barriers, old age, recall bias, and other medical comorbidities, most patients were not able to tell us if they had been vaccinated before.

Finally, logistical issues might favor the use of an IFN-γ assay in evaluating ESRD patients for TB infection. Many ESRD patients have distorted forearm architecture and skin atrophy due to disease, medical interventions (graft or shunt placement), and age. Placing and reading a TST can be challenging in these conditions. On the contrary, obtaining blood specimens is easy given the permanent vascular access in such patients. Furthermore, IFN-γ assays have obvious advantages over the TST: they can be performed in a single patient visit, they eliminate the technical difficulties with placement of a TST and the subjectivity involved in interpreting the TST, and there is no “boosting” of the immune system like that which can take place with repeated TST assessments. In our study, boosting was suggested by the fact that persons with positive TST results were more likely to have had recent previous tuberculin skin testing. No such association was observed for those with positive IFN-γ assay results.

Our data suggest that IFN-γ assays are useful in the detection of latent TB infection among renal dialysis patients. Our experience suggests they might be more sensitive and specific than the time-honored TST in diagnosing recent latent TB infection. Our study is the first to assess this technology in ESRD patients recently exposed to TB, and further studies should be undertaken. Our findings suggest that clinicians could consider using IFN-γ assays instead of TST to screen renal dialysis patients for TB infection.

Disclosures

None.