Retesting for latent tuberculosis in patients with inflammatory bowel disease treated with TNF-α inhibitors

Authors


Correspondence to:

Dr W. Reinisch, Department of Internal Medicine III, Division of Gastroenterology and Hepatology, Medical University of Vienna, Waehringerguertel 18-20, 1090 Vienna, Austria.

E-mail: walter.reinisch@meduniwien.ac.at

Summary

Background

Patients treated with TNF-α inhibitors (TNFi) are at high risk of reactivation of latent tuberculosis (LTB). Prospective studies on monitoring of TB reactivation and/or infection in this risk group are lacking.

Aim

To test the conversion and reversion rate of screening tests for latent TB serial tuberculin skin test (TST) and interferon-γ release assay (IGRA) under ongoing TNFi therapy.

Methods

We retested consecutive patients with IBD receiving TNFi therapy for a minimum of 5 months for LTB using IGRA and TST. A detailed patient history and concomitant therapy were recorded for each subject.

Results

After a median of 34.9 weeks (20.7–177.7), IGRA was retested in 184/227 patients (81.1%; Crohn's disease = 139, ulcerative colitis n = 45) still under index TNFi. TST was available in 144/184 subjects (78.2%). The majority of patients were TNFi naïve (147/184, 79.9%). In a subgroup of patients who received isoniazid due to diagnosis of latent TB at baseline (n = 32), 6/13 patients (46.2%) with baseline positive IGRA and 3/22 patients (13.6%) with baseline positive TST reverted to negative at retesting. In patients without diagnosis of LTB at baseline no permanent IGRA conversion was observed, but there were 6/144 (4.2%) TST conversions from negative to positive. No single case of TB reactivation or infection was recorded during the observation period.

Conclusions

During treatment TNF-α inhibitors conversion was observed for tuberculin skin test, but not interferon-γ release assay. As compared with tuberculin skin test, interferon-γ release assay reverted in nearly half of isoniazid-treated patients for latent tuberculosis. However, the fact that patients in whom the interferon-γ release assay test result remained positive did not develop active tuberculosis during follow-up questions the utility of interferon-γ release assay as a monitoring tool during chemoprevention.

Introduction

Patients with immune-mediated inflammatory diseases (IMID) who are treated with TNF-α inhibitors (TNFi) are at increased risk for reactivation of latent tuberculosis (LTB).[1] TNFi treatment in patients with Crohn's disease and rheumatoid disorders is associated with an approximately 14-fold increase of TB reactivation in comparison with healthy controls.[2-4] In addition, the combination of TNFi and immunomodulator (IM) therapy might further increase the risk of reactivation up to 40-fold.[3] Thus, screening for LTB is essential prior to commencing treatment with biologics.[5]

Reactivation of LTB mostly occurs during the first year of TNFi treatment,[1, 4] with a shorter median time to reactivation for infliximab (3–6 months) compared to adalimumab (8–16 months).[2, 6, 7] Furthermore, ‘de novo’ TB infection in exposed patients, e.g. travellers in endemic regions, can occur at any time during treatment.[8] The estimated risk of TB infection for healthy individuals travelling to countries with high endemic TB levels is between 2% and 12%,[9-11] but the corresponding risk for patients on biologics still remains unknown.[12, 13] A significant percentage (>10%) of IBD patients treated with biologics from our centre are immigrants from countries with higher incidence rates of TB,[14] a percentage comparable with other centres.[15] This subgroup of patients is characterised not only by a higher prevalence of LTB, but also an increased risk of acquiring new TB infection due to frequent travel to their countries of origin and their compromised immune systems.[16]

The aim of our prospective study was to test for the conversion and reversion rate of screening tests for latent TB serial IGRA and TST in patients with IBD under ongoing TNFi therapy.

Patients and Methods

Patients and study design

Inflammatory bowel disease patients were considered eligible for inclusion into this prospective cohort study if they had participated previously for routine screening for LTB at the IBD outpatient clinic at the Medical University of Vienna between December 2006 and June 2010, and if they were undergoing treatment with an anti-TNF-α agent. This study is the consecutive follow-up of a previously reported IBD cohort screening for LTB before initiation of TNFi treatment.[14] All patients were diagnosed with IBD according to the recommendations of the European Crohn's and Colitis Organisation (ECCO).[17, 18] For each patient, a detailed chart review was performed and disease-specific data were collected according to the definitions of the Montreal Classification and of a validated documentation standard for IBD (IBDIS, Inflammatory Bowel Disease Information System).[19, 20] IBD patients were treated according to international guidelines and recommendations.[21] The dose and duration of immunomodulatory therapy, including steroids, thiopurines, methotrexate, calcineurin inhibitors and TNFi at LTB screening and during TNFi were recorded.

All newborns in Austria from 1952 until late 1989 were vaccinated with BCG in the first weeks of life. For immigrant patients, BCG status was obtained from their vaccination card and/or patient information. All patients were asked for previous screenings for LTB, history and symptoms of TB, as well as risk factors for LTB, such as close contact with subjects known to have active TB, origin from countries with a higher prevalence of TB and work experience in health care facilities or nursing homes. Detailed travel history, country and duration of stay and possible exposures during TNFi treatment were also recorded at the time of retesting.

As part of routine management, each patient underwent TST and IGRA on the same day prior to the start of TNFi treatment. In case of positive IGRA, TST or chest X-ray at baseline, prophylactic therapy with isoniazid (INH) was provided for 6–9 months. After a minimum of 3 weeks of prophylaxis, biological therapy was initiated. All patients still under treatment with a TNFi for a minimum of 5 months were asked for retesting of latent TB using TST and IGRA. Starting from the end of 2010, retesting was restricted to IGRA due to a change in national guidelines.

Tests for latent tuberculosis reactivation

Interferon gamma release assay (IGRA)

A whole-blood IFN-γ release assay (QuantiFERON TB-Gold in Tube; Cellestis Limited, Carnegie, Australia) based on the Mycobacterium tuberculosis peptides ESAT-6, CFP-10 and TB7.7 was performed according to the manufacturer's instructions as described in detail previously.[14] The readers of the IGRA test in the Department of Laboratory Medicine of the Medical university of Vienna were blind to the result of the TST test and to clinical information.

Tuberculin skin test (TST)

In our centre, TST is routinely performed by trained physicians according to the Mendel–Mantoux Method. Tuberculin purified protein derivate (PPD RT23; Staten Serum Institute, Copenhagen, Denmark) at a dose of 2 tuberculin units (0.1 mL) is injected intracutaneous at the volar side of lower arm and the main diameter of the skin induration is measured in millimeters and documented 48–72 h after inoculation by an experienced examiner as reported previously.[14, 22] The readers of TST were blind to the result of IGRA test.

Chest X-ray

Findings on chest X-ray indicative of LTB included calcified granulomas, pleura scarring, apical densities and/or hilar lymphadenopathy. Follow-up chest X-ray was performed only in case of IGRA and/or TST conversion or latent TB at baseline.

IGRA/TST conversion and reversion

Interferon gamma release assay conversion was defined as a reproducibly positive result under biological therapy after a negative baseline test. TST conversion was regarded if the negative baseline TST prior to treatment start became positive (≥5 mm of induration) under biological therapy.[23] IGRA and/or TST reversion was defined as a negative test result subsequent to positive result in previous tests. All other retest results either remained stably negative or stably positive.

Definitions

Screening for LTB prior to initiation of index TNFi was regarded as the baseline assessment. Retesting was defined as follow-up evaluation during treatment with the index TNFi. LTB was defined as the presence of at least one positive result of the screening methods TST, IGRA, or chest X-ray as reported previously. The IGRA test was positive if the IFN-γ concentration was ≥0.35 IU/mL, irrespective of the result of the negative control. The IGRA test was negative if the concentration of IFN-γ was <0.35 IU/mL and the result of the positive control ≥0.5 IU/mL, and indeterminate if the result of IFN-γ concentration was <0.35 IU/mL for tuberculosis-specific antigens and <0.5 IU/mL for the positive control.[12] TST was considered positive if a diameter of induration ≥5 mm occurred in patients with IM-therapy or ≥10 mm in subjects without IM-therapy.[24]

IM-therapy was considered if steroids ≥15 mg/d were taken ≥2 weeks, thiopurines or methotrexate ≥3 months and TNFi administered during the last 12 weeks.

Statistics

For statistical analysis, spss 19 and R 2.12.2 (R Development Core Team 2011, R: A language and environment for statistical computing. The R Foundation for Statistical Computing, Vienna, Austria ISBN 3-900051-07-0) was used. Data are presented as frequencies and percentages as mean and standard deviation or median values with range respectively. To investigate the difference in the change of the proportions between baseline and retesting in IGRA and TST in baseline positive patients, 95% confidence intervals for proportions based on a binomial distribution were calculated. (significant if confidence intervals not overlapping). Differences were compared using Student′s t-test or Mann–Whitney U-test. Differences of proportions were analysed using the Chi-squared or Fisher Exact-Test. To compare the median change of IFN-γ level and TST induration in baseline positive patients with IGRA and TST under INH therapy, paired t-test was applied. Differences were considered significant if P < 0.05.

Results

Total cohort of IBD patients screened for LTB and treated with biologics

In total, 227 patients with IBD [Crohn's disease: 174/227 (76.7%)] underwent baseline screening before index TNFi. At that time, 142/227 patients (62.6%) were under immunomodulatory therapy with about 5% (11/227) being treated with a combination of immunosuppressive agents. Baseline characteristics are given in Table 1. Out of all 227 subjects in whom a TNF inhibitor therapy was started between December 2006 (when IGRA became available at our institution) and June 2010, 184 patients (81.1%) still were under treatment and available for retesting with IGRA after a median time of 34.9 weeks. TST was available in 144/184 subjects (78.2%; Figure 1). Retesting with IGRA resulted in 147/184 (79.9%) biologic naïve subjects and 37/184 (20.1%) previously treated with TNFi. Within this, re-tested cohort 152/184 (82.6%) subjects had no signs of LTB at baseline assessment, whereas 32/184 (17.4%) subjects were considered positive for LTB and were treated with INH therapy. There were 15/184 (8.2%) IGRA-positive and 27/184 (14.7%) TST positive subjects.

Figure 1.

Flow chart of patients with inflammatory bowel disease included in the study.

Table 1. Characteristics of IBD patients prior to anti-TNF-α therapy with retesting for latent TB
 All treatedIGRA retestedTNF naiveTNF experiencedP-valuea
  1. IM, immonomodulatory treatment.

  2. a

    Refers to comparison between TNFi naïve and TNFi experienced patients.

  3. b

    Data on concomitant IM-therapy were available for 178/184 (96.7%) re-tested patients.

Number (%)227184 (81.1)147 (79.9)37 (20.1) 
Age at IBD diagnosis (years ± s.d.)28.0 ± 10.828.2 ± 10.728.6 ± 10.726.4 ± 10.2N.S.
Age at screening (years ± s.d.)36.1 ± 10.636.3 ± 10.635.8 ± 11.138.2 ± 8.50.05
Time to retesting (weeks, range) 34.9 (20.7–177.7)34.8 (20.7–116)34.9 (20.7–177.7)N.S.
Women (%)118 (52)93 (50.5)75 (51)18 (48.6)N.S.
Crohn's disease174 (76.7)139 (75.5)105 (71.4)34 (91.9)0.035
IM treatment at baseline, n (%)142 (62.6)115 (62.5)87 (59.2)28 (75.7)0.06
 AZA/6-MP/MTX, n (%)112 (49.3)93 (50.5)73 (49.6)20 (54.1)N.S.
 Steroids, n (%)69 (30.4)54 (29.3)46 (31.3)8 (21.6)N.S.
 Steroids >15 mg/daily32 (14.1)22 (11.9)19 (12.9)3 (8.1)N.S.
 concurrent anti-TNF-α antibody, n (%)20 (8.8)18 (9.8)0 (0)18 (48.6) 
Double IM at baseline, n (%)15 (6.6)11(6.0)4 (2.7)7 (18.9)<0.001
IM treatment at follow-upb 56 (31.5)47 (34.5)9 (25)N.S.
Risk factor, n (%)71 (31.3)60 (32.6)53 (36.1)7 (18.9)0.047
LTB positive39 (17.2)37 (20.1)31 (21.1)6 (16.2)N.S.
 IGRA + TST positive, n (%)8 (3.5)8 (4.3)7 (4.8)1 (2.7)N.S.
 IGRA positive alone, n (%)7 (3.1)7 (3.8)6 (4.1)1 (2.7)N.S.
 TST positive alone, n (%)21 (9.3)19 (10.3)15 (10.2)4 (10.8)N.S.
 Chest X-ray positive, n (%)3 (1.3)3 (1.6)3 (2)0 (0)N.S.
INH therapy, n (%)34 (15)32 (17.4)29 (19.7)3 (8.1)0.1
Index anti-TNF-α therapy    0.014
Infliximab94 (41.4)74 (40.2)67 (45.6)7 (18.9) 
Adalimumab109 (48)92 (50)65 (44.2)27 (73) 
Certolizumab11 (4.8)9 (4.9)7 (4.8)2 (5.4) 
Golimumab13 (5.7)9 (4.9)8 (5.4)1 (2.7) 

TNFi naïve cohort

Among the 147 biologic naïve patients, the rate of LTB was 31/147 (21.1%). In 6/31 subjects (19.4%), both IGRA and TST were positive in seven patients (22.6%), whereas in 15 individuals (48.4%), only TST was positive at baseline. Another 3/31 subjects (9.7%) showed signs strongly suggestive of LTB on chest X-ray (pleura scarring and calcified granuloma), whereas IGRA and TST were negative. However, the combination of very suggestive signs of previous MTB infection with the possibility of false negative TST and IGRA assays under IM-therapy (azathioprine, methotrexate, high dose steroids) and, in two of the cases, previous contact with TB, nevertheless prompted us to classify the three patients as latently infected. As two patients declined prophylactic INH treatment, the subgroup of patients with chemoprophylaxis included 29 subjects.

TNFi experienced cohort

A 37/184 patients (20.1%) had undergone treatment with a biologic before TNFi, with the majority of these patients having been treated previously with adalimumab. The rate of LTB was 6/37 patients (16.2%), three of whom received INH prophylaxis (two due to TST conversion 5 and 7 years after a previously negative TST and one due to a positive IGRA result after a previously negative TST, but when IGRA had been not available). Another three TNFi experienced patients with documented INH therapy for 9 months prior to first TNFi agent were regarded as sufficiently treated for LTB and retreatment was not provided.

Patients with LTB were significantly older, had significantly more risk factors for TB infection and had previously received significantly less biological therapy as compared with those without LTB.[14]

Reversion of TST and IGRA under TNFi therapy and prophylactic therapy with isoniazid

In 5/27 (18.5%) initially TST positive patients and in 2/15 (13.3%) initially IGRA-positive subjects, no retesting was performed. Chemoprophylaxis for LTB with isoniazid (INH) 300 mg daily for median of 8 (range 6–9) months was provided concomitantly to biological therapy in 32 patients. After a median of 28 weeks (20.4–83.3) from the start of TNFi therapy, IGRA reversion occurred in 6/13 (46.2%) patients, whereas TST reverted under INH treatment in 3/22 (13.6%) re-tested patients (Figure 2). In 1/13 subject (7.7%), the serial IGRA was indeterminate.

Figure 2.

Proportion of patients with positive tuberculin skin test (TST) and interferon-gamma release assay (IGRA) in patients with latent TB treated with isoniazid at baseline (BL) and follow-up (FU).

The corresponding change of proportions based on binomial distribution was 46.15 (19.22–74.87) for IGRA and 13.60 (2.66–33.86) for TST (not significant). The median IFN-γ level in 13 IGRA-positive subjects dropped from baseline 1.04 (0.43–9.41) IU/mL to 0.24 (0.0–11.2) IU/mL at retesting (= 0.57).

None of the patients with a positive TST, but a negative IGRA at baseline who received prophylaxis with isoniazid became IGRA-positive during retesting, whereas two patients with previously negative TST became TST positive under INH (one IGRA-positive at baseline and one chest X-ray positive each at baseline). The median TST induration in TST positive patients was 13.5 (5–28) mm at baseline and 12.5 (0–21) mm at follow-up (= 0.46).

Conversion of TST and IGRA in patients under TNFi therapy

There was no case of permanent IGRA conversion in the group of subjects without LTB at baseline after a median of 35.7 weeks (20.7–177) from start of TNFi therapy. However, there were four obviously unspecific IGRA variations close to the cut-off point for positivity (0.2–0.7 IU/mL), which were not reproducible by further IGRA measurements. In 6/144 (4.2%) patients with a previously negative TST at baseline, a conversion of TST was observed after a median of 33 weeks (20–116) from therapy start (Table 2). Until October 2011, none of the patients developed active TB during a median follow-up 39 (14–91) months.

Table 2. Conversion of TST
Age/sexTB historyIBDBaselineRetestingMonths to conversionMonths to follow-up
TST (mm)/IGRA (IU/mL)IM-therapyTST (mm)/IGRA (IU/mL)IM-therapy
  1. AS, ankylosing spondylitis; CD, Crohn's disease; UC, ulcerative colitis.

41/mTB contact with 13yUC + AS0/indeterminateAzathioprine5/0.0Infliximab933
22/mNo specific historyCD0/0.00No7/0.00Infliximab750
31/fNo specific historyCD0/0.05Azathioprine6/0.00Adalimumab844
25/fOrigin from BosniaCD0/0.00Azathioprine8/0.00Infliximab746
46/fMedical staff, suspected contactCD3/0.00Steroids 5 mg10/0.00Adalimumab1624
41/mNo specific historyCD0/0.03Budesonide 9 mg8/0.03Adalimumab814

Discussion

TNF-α inhibitor therapy in patients with IMIDs increases the risk of reactivation of latent tuberculosis by up to 40-fold.[3] Despite this increased risk, studies prospectively monitoring TB reactivation by screening tests in this population are scarce.[13] In this study, we present the first comparison of serial IGRA and TST in long-term monitoring of LTB in IBD patients treated with biologics.

Of the approximately 800 patients who had started treatment with biologics in our institute since 1999, we did not observe a single case of TB reactivation or infection by November 2011 (data not shown), including for the 184 subjects in this study. Among those patients without LTB at baseline, we observed six conversions of TST. This finding might be best explained by a booster reaction and/or a lower specificity of TST, a well-known weakness of TST, as the corresponding IGRA results were negative. Furthermore, the fact that all patients included had undergone BCG vaccination during childhood might have resulted in a higher rate of baseline TST positivity in our cohort. This could have also impacted TST conversions. All the subjects with TST conversion had a negative chest X-ray and none of those received prophylactic treatment with isoniazid after consulting an infectious disease specialist. Interestingly, there were four subjects with marginally positive results from IGRA at retesting, although further repetitions of the test revealed negative values. This observation is consistent with little published data on this topic, where within-subject variability occurred in up to 25% of subjects, regardless of low or high incidence countries.[25] These four subjects were therefore considered to have negative results in the IGRA retesting.

There are only a few studies focusing on TB reactivation in patients with IMIDs under TNFi treatment.[13, 27] A study from Taiwan (a country of high TB incidence) on 43 patients treated with adalimumab for rheumatoid arthritis observed conversion of TST in 10/27 (37%) subjects during therapy. At the time of retesting, two patients with TST conversion had a positive IGRA result and one had developed active TB disease. Among the 17 patients without TST conversion after 12-month adalimumab therapy, one patient who had a positive IGRA result at retesting developed active TB disease subsequently. Thus, in that study, IGRA appeared to be both more sensitive and specific in the prediction of active TB then TST.[26] In a retrospective, observational study from the United States (a country of low TB incidence), 58% (n = 198) of all patients were tested with IGRA, while on TNFi therapy.[13] The authors observed a 0.3% rate of active TB during a follow-up of 1.5 years. The single patient with TB reactivation was previously anergic for TST, whereas IGRA was indeterminate under treatment with infliximab. Repeated IGRAs performed in 131 patients who previously had either negative or indeterminate results were identical in 97.4% of subjects. However, direct comparison of our results with this study is limited because the majority of patients in that study was tested with IGRA while on maintenance therapy with TNFi and test-retest results prior to and during treatment were not reported. A recent report by Bermejo et al., published only in abstract form, found almost identical results, with a rate of TST conversion of 3% (using the same definition of TST conversion) and no IGRA conversion in 67 IBD patients under TNFi treatment.[28]

Among patients treated with isoniazid for latent TB at baseline, we observed a higher rate of reversions to normal at retesting for IGRA vs. TST results. In addition, there was a quantitative decrease in IFN-γ levels. The fact that our results were not statistically significant might mainly derive from sample size limitations. Our observation of a treatment-associated IGRA reversion rate of 46.1% is consistent with previous reports on a decline in IFN-γ release as well as reversion in IGRA assays on patients with smear-positive TB infections under treatment for active TB, with reported ranges from 31.6 to 39.2%.[29-31] Previous reports from non-IBD populations likewise show a decline of IFN-γ responses during INH treatment.[32-34] Out of the 13 subjects with positive baseline IGRA under INH therapy, the serial IGRA became indeterminate in one patient (7.7%), which is consistent with our previously published data.[36] When interpreting the decline and reversion of serial IGRA, the highly dynamic test results due to fluctuations of T-cell responses over time also need to be considered.[25] This and the fact that patients in whom the IGRA test result remained positive did not develop active TB during follow-up; however, questions the utility of IGRA as a monitoring tool during chemoprevention.

The majority of our patients who reverted had a higher burden of immunomodulatory therapy at baseline compared with follow-up. As immunosuppressive therapy, in particular, steroid treatment, negatively impacts the results of IGRA, it appears unlikely that reversion in our patients is an artefact arising from concomitant immunosuppression. However, the direct impact of TNF inhibitors cannot be excluded as a reduction of IFN-γ release was observed under combination of anti-TNF agents and immunosuppressants.[14, 35, 37]

As no cases of TB were detected in our cohort, estimates of the sensitivity and specificity of IGRA and TST during retesting cannot be derived. According to a recent meta-analysis, the reported specificity for IGRA is 98–100% with a pooled specificity of 99.4% and for TST is 55–95% with a pooled specificity of 88.7%.[12] However, these numbers are mainly based on subjects with close contact to TB cases or HIV infected patients with very low CD4+ lymphocytes counts, as large comparable studies are missing in patients with IMIDs under TNFi therapy.[38, 39]

As already mentioned, the major limitation of our and most other studies on LTB testing is the number of patients included and followed up. On 19% of subjects originally screened for LTB at baseline, no retesting was available. However, baseline characteristics between our total cohort and the patients with retesting were balanced. Our reported drop-out rate during the course of TNFi therapy is rather in the lower range of published literature. From the randomised controlled trials with infliximab or adalimumab in IBD, drop-out rates of up to 40.6% are known from a 12-month follow-up.[40-43]

In conclusion, our results show that IGRA reverts in nearly half of isoniazid treated patients as compared with TST. IGRA seems to be a more responsive method to monitor BCG-vaccinated patients under treatment with TNFi than TST. However, the utility of IGRA as a monitoring tool during chemoprevention remains questionable as the clinical relevance of non-reversion remains to be determined. Furthermore, false, marginally positive results from IGRA deserve careful attention by treating physicians. Nevertheless, IGRA is moving forward to become the primary test to reduce the risk of latent TB reactivation and to optimise the management of patients with IBD.[5, 44, 45] Since the end of 2010, new guidelines on LTB screening have been established in Austria omitting TST as a screening test. According to these recommendations, only 42% of the originally treated patients would have received chemoprophylaxis with isoniazid.[2, 4]

Acknowledgements

Declaration of personal interests: We thank David Berry for critical reading of the manuscript. Declaration of funding interests: None.

Ancillary