The opinions are solely those of the authors and do not represent an endorsement by the Department of Defense or the National Institutes of Health. This is a US Government work. There are no restrictions on its use.
The incidence, risk factors, and prognosis for Mycobacterium tuberculosis (MTB) infection have not been reported in a national population of renal transplant recipients. We performed a retrospective cohort study of 15 870 Medicare patients who received renal transplants from January 1 1998 to July 31 2000. Cox regression analysis derived adjusted hazard ratios (AHR) for factors associated with a diagnosis of MTB infection (by Medicare Institutional Claims) and the association of MTB infection with survival. There were 66 renal transplant recipients diagnosed with tuberculosis infection after transplant (2.5 cases per 1000 person years at risk, with some falling off of cases over time). The most common diagnosis was pulmonary TB (41 cases). In Cox regression analysis, only systemic lupus erythematosus (SLE) was independently associated with TB. Mortality after TB was diagnosed was 23% at 1 year, which was significantly higher than in renal transplant recipients without TB (AHR, 4.13, 95% CI, 2.21, 7.71, p < 0.001). Although uncommon, MTB infection is associated with a substantially increased risk of mortality after renal transplantation. High-risk groups, particularly those with SLE prior to transplant, might benefit from intensified screening.
It is well reported in the worldwide transplant community literature that end-stage renal disease, dialysis, and renal transplant patients are at an increased risk for both primary and reactivation Mycobacterium tuberculosis (MTB) infection compared with the general population. The decreased cellular immunity of patients with end-stage renal disease, secondary to uremia, and the iatrogenic immunocompromised state post-transplant are commonly believed to be contributors. This paper focuses on patients who have undergone renal transplantation in the US, looking at incidence of Mycobacterial infection after renal transplant compared with the general US population and worldwide incidence. Current practices on screening and treatment of MTB infection are then examined.
In 1983, Spence found that the incidence of primary MTB infection after renal transplantation in the United States was approximately 0.5% (1). That study reviewed 565 renal transplant recipients, of whom three developed primary MTB infection after transplant; two others developed other Mycobacterial infection. That study attempted to focus on primary MTB infection, although there was no comment on other patients who may have developed reactivation infection. The cumulative incidence of MTB infection in the renal transplant community has not been examined on a larger scale in the US since the study in 1983.
Data were obtained from the United States Renal Data System (USRDS), using standard analysis files (SAF) as of October 2001. The variables included in the USRDS SAF, as well as data collection methods and validation studies and demographics of the population, are listed at the USRDS website, under ‘Researcher's Guide to the USRDS Database', Section E, ‘Contents of all the SAF's’, http://www.usrds.org and published in the USRDS. SAF.TXUNOS was used as the primary dataset, and merged with variables from SAF.TXFUUNOS for follow-up data, Institutional Claims Details for Medicare Claims data, and SAF.PATIENTS for dates and causes of death as well as causes of renal disease, as previously reported (2,3). Because our outcome was a Medicare-defined event, patients were limited to those who were identified as having Medicare as a primary payer as defined in previous studies (4). No age ranges were excluded, thus including pediatric patients so long as they were eligible for Medicare at the time of transplant.
Our main outcome was Medicare Claims with a diagnosis of M. tuberculosis[TB, based on International Classification of Diseases-9th Modification Diagnosis Codes (ICD9) codes 01x.x, excluding tuberculosis in pregnancy (647.x), tuberculosis contact (V011.x), vaccine for tuberculosis (V032.x), personal history of tuberculosis (V1201), or late effect of tuberculosis(137.x)]. We assessed the first Medicare Claim for TB after renal transplantation for a given individual who received a kidney transplant occurring on or after January 1 1998 and before July 31 2000 (which could have been a repeat transplant, but only one transplant was assessed per patient during the time-period of the study). Follow-up time was censored at 3 years. We also assessed patient survival after TB.
Variables used in analysis
Independent associations between patient factors and Medicare Claims for TB were examined adjusting for recipient and donor age, recipient race, gender, weight, human leukocyte antigen (HLA) mismatch, peak panel reactive antibody percentage, cold ischemic time, pre-transplant dialysis (yes/no), duration of dialysis prior to transplantation, total follow-up time, repeat transplant, donor and recipient cytomegalovirus and hepatitis C serology, recipient human immunodeficiency virus (HIV) status, dialysis in the first week after transplant (delayed graft function, yes/no), rejection (either treatment or diagnosis) occurring at any time in the study period, induction antibody therapy, induction sirolimus therapy, maintenance immunosuppressive medications at time of discharge after transplant surgery, and cause of ESRD (assessed as diabetes, systemic lupus erythematosus, and all others). Induction/maintenance immunosuppressive medication use at the time of discharge after transplant was analyzed as a pre-existing covariate. Previous dialysis modality was obtained from the file SAF.RXHIST60.
Time to TB was defined as the time from first renal transplant until the date of claim for TB, with patients censored at death, loss to follow-up, 3 years after the date of transplant, or end of the study (which was considered December 31 2000, the most recent date of Medicare Claims available). Thus, for time to TB, follow-up time did not continue to accrue after the date of diagnosis of TB. Survival time to death was calculated as the time from the date of transplant until the date of death, censored for the end of the study period or loss to follow-up.
All analyses were performed using SPSS 12.0 TM (SPSS, Inc., Chicago, IL, USA). Files were merged and converted to SPSS files using DBMS/Copy (Conceptual Software, Houston, TX, USA). Univariate analysis was performed with Chi-square testing for categorical variables (Fisher's exact test for violations of Cochran's assumptions) and Student's two-sided t-test for continuous variables (Mann–Whitney U-test used for non-normally distributed variables). Variables with p < 0.10 in univariate analysis for a relationship with development of TB were entered into multivariate analysis as covariates. Kaplan–Meier analysis was used to construct survival plots of time to TB after renal transplantation. Log-minus-log plots were used to assess the validity of the proportional hazards assumption over time. Stepwise Cox regression (likelihood ratio method) was used to model factors associated with time to TB, controlling for covariates listed above. Stepwise logistic regression was used to model factors associated with evidence of Medicare as a primary payer, adjusted for demographic- and transplant-related variables that might influence the generalizability of results of the analysis.
Of 34 920 patients who received renal transplants in the USRDS database from January 1 1998 to July 31 2000, 32 779 had information complete enough to calculate survival times, of whom 16 257 had documentation of Medicare as their primary payer at the time of transplant. Of these, 15 870 (97.6%) had a Medicare Claim filed within 14 days of transplant indicating Medicare as the primary payer. The descriptive characteristics of this cohort, including associations with Medicare as primary payer, have been reported previously (4).
Of the study population, 66 (0.4%) had Medicare Claims for TB after renal transplantation during the time period of the study, with a mean follow-up of 1.63 ± 0.8 years, yielding an incidence density of TB of 2.48/1000 person years (95% CI, 1.94, 3.17). Time to TB after renal transplantation is shown in Figure 1. The rate of TB was a relatively constant rate over time, with possible falling off of cases with time (0.31% at 1 year).
The detailed diagnoses of patients with TB are shown in Table 1. The most common diagnosis was pulmonary TB (41 cases, 62%). There were eight cases (12%) of TB limited to the skin, four (6%) cases of military TB, followed by other diverse sites of TB.
Table 1. Sites of TB infection
Site of TB Infection
Pulm TB NOS – unspecific
Skin – subcutanueous (scrofula)
TB peripheral lymph
In univariate analysis, only end-stage renal disease due to systemic lupus erythematosus [SLE, n = 513 (3.7% of all recipients), of whom 8 (1.6%) had had TB compared with 0.4% of all other patients, unadjusted odds ratio, 4.07, 95% CI, 1.93–8.64], graft loss (unadjusted odds ratio, 4.12, 95% CI 2.27, 7.45), and rejection occurring in the first year (unadjusted odds ratio, 2.21, 95% CI, 1.32, 3.69) after transplant were significantly associated with TB. Specifically, no immunosuppressive medications, whether as induction or prescribed at the time of discharge immediately after the transplant surgery, were associated with TB. Neither were any dialysis or transplant-related factors, including diabetes, donor type, the duration of dialysis prior to transplant, or viral serologies (CMV, HCV, or HIV, whether donor or recipient). When limited to patients who survived the first year, allograft rejection occurring in the first year after transplant was not associated with TB, p = 0.17, although the trend was still increased, univariate odds ratio, 1.99, 95% CI 0.74, 5.35.
In multivariate analysis by Cox regression, adjusting for donor and recipient age, race, gender, donor type, diabetes, duration of dialysis prior to transplant, and type of immunosuppressive medication at discharge, only SLE (adjusted hazard ratio, 4.14, 95% CI, 1.82, 9.41, p < 0.001) was independently associated with TB.
Figure 2 shows case fatality after TB, which was particularly high early after the date of the Medicare Claim for TB. Specifically, mortality 1 year after diagnosis of TB was 23%, significantly higher than for comparable renal transplant recipients (p < 0.01 by Log Rank test). The significance of this association persisted in adjusted analysis; as a time-dependent variable in Cox Regression, TB was independently associated with death after renal transplantation, adjusted hazard ratio, 4.13, 95% CI, 2.21, 7.71, p < 0.001. However, TB was not significantly associated with graft loss, p = 0.58, adjusted hazard ratio, 1.75, 95% CI, 0.24, 12.41. Among patients with TB who died, 50% had valid specified causes of death. There was one death each recorded for cardiac arrhythmia, sepsis due to infected vascular access, sepsis unspecified, fungal pulmonary infection, unspecified pulmonary infection, tuberculosis, and liver failure.
The present study indicates that while MTB remains the most frequent cause of death from infection in the world (5), it is a fortunately rare occurrence among recent US renal transplant recipients, occurring in 0.3% of recipients in the first year with an incidence of 248/100 000 person years. Published reports from around the world list the incidence of MTB infections in renal transplant patients ranging from 0.5% to 14.7%. According to the 1998 National Hospital Discharge Survey, there were 19 000 cases of tuberculosis as an all-listed diagnosis in the United States, corresponding to an annual rate of 6.95/100 000. Because previous reports of transplant associated MTB did not account for potential years at risk, comparison of incidence rates are not possible. Not all of these reports distinguish primary from reactivation cases. The wide range in percentage is likely due to the different endemic rates of MTB and/or varying practice in screening and providing prophylaxis with isoniazid (INH). Center-specific experience with drug toxicity and prevalence of liver disease are factors that limit the use of INH prophylaxis. However, although MTB was rare among US renal transplant recipients, it was associated with a high risk of mortality. Causes of death were consistent with infection.
The only factor independently associated with the development of post-transplant MTB infection was SLE. Notably, the type of immunosuppressive medications used, specifically use of induction methylprednisolone or antibody therapy, was not associated with the development of MTB. Although we were unable to determine previous use of corticosteroids, results or type of purified protein derivative (PPD) screening, or use of prophylactic drug therapy for MTB, this suggests that patients with prolonged or high-dose corticosteroid use prior to renal transplantation are at high risk of post-transplant MTB, which is associated with poor survival. Alternatively, other factors associated with SLE, possibly immune surveillance defects unique to these patients, consistent with the increased risk of patients with SLE for infection even in the absence of immunosuppressive therapy (6,7), may be associated with the increased risk of post-transplant TB.
In the US, as part of the pre-transplant workup, recommendations are that patients be interviewed at length for risk factors for TB (e.g. TB exposure history, prior residence in an endemic area, malnutrition, prior TB skin testing, high-risk population, other immunocompromising conditions, or prior chest X-ray (CXR) findings suggestive of MTB) (8). Some of these factors are associated with having latent TB (prior residence in endemic areas, for example) while others predispose to reactivation (i.e. malnutrition), particularly once patients undergo the immunosuppression associated with the post-transplant medications. All patients also undergo CXR and PPD skin testing (if no history of positive PPD in the past) (8).
There is no current standard of care with regard to INH prophylaxis in the US transplant community because there are no adequately powered, randomized controlled trials to determine the efficacy of treatment (9). The American Thoracic Society and the Centers for Disease Control and Prevention have written in their guidelines that ‘Nine months of INH (and pyridoxine) is recommended for kidney transplant candidates and recipients who have never received adequate treatment and who are tuberculin test positive, have a history of tuberculosis, have a CXR suggesting latent TB, have a recent exposure history, or have received a kidney from a TB test positive donor’. Further, patients should receive this therapy prior to transplant. Even though the American Thoracic Society and the Centers for Disease Control and Prevention jointly developed guidelines for the treatment of latent MTB in 2000 (10), it is still up to the local authority to decide 6- vs. 9-month therapy in the US.
Among screening tests, almost all countries use the standard Mantoux skin test for MTB identification. The standard reaction size is also universal at 10 mm for non-HIV-infected, dialyzed patients regardless of BCG vaccine status (11,12). A 5 mm induration standard after 48–72 h is used for patients who are HIV-positive, have a CXR with findings consistent with healed tuberculosis, and/or known close contact with an MTB-infected individual. Recent reviews have proposed this same standard for non-HIV positive individuals who are immunosuppressed (13). However, a recent report indicates that skin testing (even with 5 mm induration as the standard) significantly underestimates in vitro PPD reactivity (85% of control patients with a positive skin test had in vitro PPD reactivity compared with 51.4% among hemodialysis patients, p = 0.007) (14). This report also suggested that in vitro quantitation of PPD-specific T cells using a whole blood assay was relatively unaffected by uremia-associated immunosuppression, and might also be useful in distinguishing between PPD reactivity due to MTB infection vs. BCG-induced reactivity. However, the potential role for in vitro PPD assays in pre-transplant screening has not been investigated.
The reason donors and recipients are tested is because the mortality risk for under treated or untreated active MTB in the renal transplant patient nears 30% (over a time period not specified consistently) (15,16), which is comparable to the 1-year mortality of patients with TB infection in the present study. Patients with a history of MTB infection should have documented whether or not they underwent therapy and if so, the medications and duration of that therapy. If the patient has a positive PPD, most centers require that they undergo pre-transplant and/or post-transplant INH therapy if they have no contraindications to that treatment. Even with a negative PPD, many centers worldwide now consider INH prophylaxis for MTB infection for patients with risk factors because of the number of end-stage renal disease and dialysis patients who are anergic (16). Unfortunately, under this definition, all patients with end-stage renal failure could be considered ‘high-risk’. Anergy can only be determined by a full anergy panel, in addition to the use of the standard PPD. It is not known how effective use of anergy panels would be in the pre-transplant screening process. In any case, recommendations for use of PPD with anergy panels in long-term dialysis patients (17), specifically for pre-transplant screening, are inconsistent (18), and the proportion of patients who actually receive PPD with anergy panels is not known (19). Among patients with rheumatic diseases, in particular SLE, extrapulmonary TB is more common and may pose diagnostic challenges as the PPD is often falsely negative in such patients, regardless of corticosteroid use (20,21). The behavior of TB among SLE patients with severe chronic kidney disease, including those on dialysis, has not been described but would presumably be even more challenging. In addition, given the high prevalence of liver disease in long-term dialysis patients, primarily due to hepatitis C infection (21), empiric use of INH therapy may pose a greater risk than in the general population.
There is also the perception that the renal transplant population is at higher risk of hepatotoxicity from INH therapy over the general population. This perception is likely based on a report by Singh et al. in 1998. That study reported significant drug-induced hepatotoxicity, requiring discontinuation of INH in 5/198 (2.5%) renal transplant patients (13). Conversely, a study by Antony et al. in 1997 found no statistically significant hepatotoxicity in the renal transplant population over the general population taking INH (actual duration of therapy almost 12 months). That study evaluated 83 renal transplant recipients on INH therapy (eight of whom had chronic Hepatitis B or C infection) and none had to have INH therapy discontinued (22). The benefits in preventing reactivation are clear in those at high risk of reactivation MTB, but not as clear in the renal transplant population at large.
Worldwide, particularly because of the high mortality associated with MTB infection post-transplant, it is the policy of many renal transplant units to give 12 months of INH therapy to patients felt to be high risk of reactivation (23) even in the absence of a positive PPD. However, in 2002, the European best practice guidelines for renal transplantation were released recommending 9 months of INH for patients with latent MTB (24).
If active MTB infection post-transplant develops, a 2–4 drug anti-tuberculosis regimen then becomes necessary; this regimen can be life saving, although it carries significant morbidity. The variation in morbidity of anti-tuberculosis therapy depends in part on the immunosuppressive regimens. Patients on cyclosporine A (CsA) (23,25–27) and tacrolimus (28,29) seem to be at higher risk of graft loss when these medications are combined with an anti-tuberculosis drug such as rifampin. This is in part due to the cytochrome P450 induction of the metabolism of CsA by rifampin and the need for increased dosing of CsA.
The present study has several limitations, consistent with previous registry studies (4). Our data likely underestimates the actual number of cases because it is based solely on hospital discharge summaries. It is also likely that there are numerous cases of MTB treated in the outpatient setting that are not reported to this database, which might have the effect of lowering mortality post-TB compared with what we report. We were unable to determine how patients were screened or treated prior to transplant. However, many other studies have found that pulmonary TB is the more common post-transplant disease, consistent with the findings of the present study, and it is likely that the incidence of TB among US renal transplant recipients truly has declined over time. Although the incidence of TB among US renal transplant recipients was lower in the present study than in previous studies, case fatality associated with TB unfortunately was not.
This retrospective cohort study supports the belief that renal transplant patients in the United States are at higher risk of either primary or reactivation M. tuberculosis infection (0.3% at 1 year, or 248/100 000 person years) compared with the general population (6.95/100 000 annual rate), although this incidence is lower than in previous reports of renal transplant-associated tuberculosis infection, consistent with reductions in the rates of TB in the general populations of developed countries. Despite this, and despite presumably careful screening during the pre-transplant evaluation process, MTB infection was associated with a greatly increased risk of death after renal transplantation. The only factor independently associated with the development of MTB infection was SLE. Whether this is a marker for pre-transplant use of corticosteroids, which may adversely influence the interpretation of the most commonly used screening tests, or other immune defects specific to SLE could not be determined. In the meantime, patients with SLE should be regarded at increased risk for the development of post-transplant MTB. Because of the increased risk of empiric anti-tuberculous therapy among renal transplant recipients, further studies of screening for MTB in high-risk populations are warranted. It would also be useful if future studies of the incidence of TB in transplant populations use either actuarial rates or rates per patient years of follow-up to facilitate comparison of studies.
Special thanks to Ms. Joyce Hershey for her editing and critical eye of the material.