Tuberculosis in the cytokine era: What rheumatologists need to know


Anticytokine agents and other biologic response modifiers represent an important advance in the treatment of rheumatoid arthritis (RA) and other inflammatory diseases. While these agents are specific in their targets, they are not precise enough in their actions to avoid the negative consequences of perturbing the immune system. In this issue of Arthritis & Rheumatism, Gómez-Reino and colleagues report findings from a clinical database initiated in February 2000 to assess the safety of tumor necrosis factor (TNF) inhibitors used to treat patients with rheumatic diseases (1). The investigators compared infection rates in two cohorts of RA patients: those who had and those who had not received either etanercept or infliximab, the two TNF blockers in use during the time period studied. They found 17 culture-confirmed cases of tuberculosis (TB) among their cohort of 1,540 patients receiving TNF blockers. The authors calculated extraordinary incidences associated with the use of infliximab: 1,893 TB cases per 100,000 patients in the year 2000 and 1,113 TB cases per 100,000 patients in the year 2001. In contrast, the incidence of TB was 21 cases per 100,000 inhabitants in Spain in the year 2000, and there were 5.8 TB cases per 100,000 inhabitants in the US in the same year (2).

Gómez-Reino et al used data from an active surveillance project specifically designed for long-term followup to assess the safety of biologic response modifiers in patients with rheumatic disease in Spain. The Spanish registry is a significant step ahead of passive postmarketing surveillance programs and is surely welcome in the field of pharmacoepidemiology. However, the authors do not report on the “capture rate” or completeness of the database. In other words, what proportion of patients receiving the agents was actually registered in the database? Without this information, we are uncertain as to the completeness of the denominator, and therefore, the extremely high calculated TB case rates must be viewed with some uncertainty.

The report, which characterizes the 17 patients with TB, demonstrates the severity of TB associated with the use of TNF inhibitors. In this study, 6 of 17 patients (35%) developed disseminated or hepatosplenic disease, and 5 patients (29%) were reported to have disease affecting the nervous system. Sixty-five percent of patients had extrapulmonary TB, compared with the usual distribution of 70–80% with pulmonary disease and 15–25% with extrapulmonary disease in community populations in Europe and the US (2). Only patients with advanced acquired immunodeficiency syndrome (AIDS) are reported to have similarly high rates of extrapulmonary disease. In addition, 2 of 17 infliximab-treated patients (12%) died, compared with a TB mortality rate of 4.6% in the US in the year 2000.

The experience in Spain thus illustrates the association of infliximab treatment with the development of active TB in persons with latent TB. Animal models have suggested this possibility, since TNF plays a key role in granuloma formation, an essential element for host control of mycobacterial infections (3, 4). In humans, a role for TNF has not been as easy to delineate, whereas inherited interferon-γ (5) and interleukin-12 (IL-12) receptor abnormalities (6) have been associated with severe mycobacterial infections. Nevertheless, between 1998, when the US Food and Drug Administration (FDA) licensed the TNF inhibitor infliximab (Remicade; Centocor, Malvern, PA) and May 2001, 70 cases of TB were reported to the FDA's Adverse Event Reporting System. The cases were described in detail by Keane and colleagues (7).

As of March 2003, there have been 242 cases of TB associated with infliximab. No cases of TB were reported during early clinical trials of etanercept. Postmarketing data thus far include a few cases of TB, but not more than would be expected according to country-specific TB incidence rates (online at The March 4, 2003 FDA safety report also includes data on the newest licensed anti-TNF agent, adalimumab (Humira; Abbot Laboratories, Abbot Park, IL). Eighteen cases of TB have been reported among 2,334 patients in premarketing placebo-controlled studies, with no TB-related deaths reported.

Since rheumatologists are likely to be confronted with TB-related issues, the state of the art for diagnosing and treating latent and active TB will be reviewed in this editorial. Diagnosing and treating latent or active TB can be complicated because of overlapping medication toxicities, potent drug interactions with rifampin (RIF), a mainstay of TB therapy, and unusual extrapulmonary manifestations of disease in this population. It is best to consult an infectious disease specialist, pulmonologist, or other TB expert early in the process. Table 1 lists and defines several of the TB-related terms discussed below.

Table 1. Tuberculosis (TB)–related terms discussed in this editorial
Latent TB infection (LTBI; also called inactive, remote, or old TB)Occurs when a person has been exposed to enough TB bacilli to elicit an immune response, as evidenced by delayed-type hypersensitivity (DTH) reaction when purified protein derivative (PPD) is injected intradermally. A positive PPD test result in the absence of symptoms or chest radiograph evidence of disease constitutes LTBI.
TB disease (also called active TB disease, TB, consumption)Occurs when a person has progressed from TB exposure or LTBI to active replication of bacteria, causing either symptoms or infiltrate on chest radiographs. Pulmonary TB is the most common kind of TB in the world, but acquired immunodeficiency syndrome and biologic response modifiers increase the likelihood of extrapulmonary disease, with or without concomitant pulmonary disease.
Contagiousness of TBPulmonary TB is primarily the form of TB that is contagious. Contagiousness depends on the concentration of organisms in the airways, the effectiveness of aerosolizing the bacteria by coughing, the concentration of the organisms in the air, and the length of time the air is breathed by susceptible hosts. An acid-fast bacilli (AFB) sputum smear–positive adult coughing in a small, poorly ventilated space is more likely to infect his or her roommates than is a child with a less-effective cough.
Tuberculin skin testing (also called PPD test, TB skin test)Intradermal injection of 0.1 cc of 5 tuberculin units of standardized PPD is the preferred method of testing and has replaced other tests in the US. Heaf and Tine tests are not recommended due to lack of uniformity of administration or interpretation. The amount of induration (not erythema) is measured 48–72 hours after injection (any earlier reactions do not represent DTH).
Mycobacterium tuberculosis (Mtb; also called AFB, “red snappers,” Koch's bacilli)The bacterium that causes TB. There are 3 species in the group called Mtb complex, including Mtb, M bovis, and M africanum. Mtb causes >98% of TB disease, but the other agents in the group can cause disease identical to TB, although such disease is less infectious to others. M bovis can infect cattle and can then be spread in their milk. M bovis is typically resistant to pyrazinamide, 1 of the 4 main TB drugs.
AFB positiveAll mycobacteria will stain positively with Ziehl-Neelsen and similar dyes; thus, a specimen that is AFB positive may be Mtb or other nontuberculous mycobacteria. Mtb is identified when it begins growing in either liquid or solid media. A biologic specimen sent to the microbiology laboratory for routine culture will not be examined in such a way as to identify AFB unless testing is requested. Most laboratories will understand if the order reads “rule out TB” or “AFB.”
Treatment of LTBI (also called preventive TB therapy, TB chemoprophylaxis, TB prophylaxis)Once a PPD test has been identified as positive and the person has been screened for symptoms of TB and found to have clear lungs on chest radiographs, treatment of LTBI can be offered.

Natural history of TB

TB is caused by an aerobic, slow-growing mycobacterium, Mycobacterium tuberculosis (Mtb). The bacteria are most efficiently transmitted when they are growing exuberantly in a lung cavity that communicates with a bronchus, causing irritation and cough. Coughing effectively aerosolizes droplets that carry the infection into the alveolar space of the new host. The likelihood that a person with TB will infect someone else is related to the number of organisms being expelled (i.e., acid-fast bacilli [AFB] smear–positive cases are most infectious), the frequency of expulsion (i.e., coughing frequency and strength), and the time that a susceptible host is breathing shared air. The susceptibility of the host is also an important determinant of infection.

After exposure to TB, the next 12 months represent a critical time when an individual's cell-mediated immunity does or does not contain the infection. If an effective exposure occurred, a person's tuberculin skin test (TST) will produce positive results, as indicated by induration measured 48–72 hours after placement. A person found to be TST negative when tested soon after a recent exposure and who demonstrates induration when retested in 3–12 months is considered a new “converter,” one of the higher risk categories for progression to active disease. However, someone who has not been tested for years and who is found to have induration on testing for reasons unrelated to a recent exposure may be at very low risk for developing active disease, unless medical comorbidities intervene, lowering immune defenses.

TB can exist for years as a latent infection, and ∼90% of people of northern European descent who have latent TB infection (LTBI) never manifest active disease. The conditions most closely associated with an increased likelihood that LTBI will progress to TB disease are listed in Tables 2 and 3.

Table 2. Incidence of active TB in persons with positive results on PPD tests, by risk factor (see ref.9)*
Risk factorTB cases/1,000 person-years
  • *

    TB = tuberculosis; PPD = purified protein derivative; HIV = human immunodeficiency virus.

Recent TB infection of <1 year12.9
TB infection of 1–7 years1.6
HIV infection35–162
Underweight by >15%2.6
Radiographic findings of prior TB2.0–13.6
Table 3. Risk for developing active TB relative to control population, independent of PPD test status (see ref.9)*
Clinical conditionRelative risk
  • *

    TB = tuberculosis; PPD = purified protein derivative.

Diabetes mellitus2.0–4.1
Chronic renal failure10.0–25.3
Renal transplant37
Carcinoma of head or neck16

Preventing LTBI from progressing to active TB

A careful history is critical in evaluating the possibility of latent TB, with queries focusing on whether the person has had a likely exposure to TB (Table 4). LTBI is routinely diagnosed by measuring the area of induration resulting from a properly placed TST. Anyone placing TSTs should be trained properly, so that the antigen, PPD, is injected intradermally (not subcutaneously), forming a tense bleb when completed. Some persons will have a wheal-and-flare reaction in the minutes or hours following placement. These reactions are not related to the delayed-type hypersensitivity response being evaluated for TB infection and should be ignored. Between 48 and 72 hours, a trained observer palpates the TST site, ignoring any erythema. If there is induration at the site, it should be measured across the width of the arm and recorded in millimeters.

Table 4. Suggested questions and observations about tuberculosis (TB) exposure
Have you ever been in jail, prison, or detention? Lived in a shelter? Injected drugs or smoked crack cocaine?
Has anyone ever said you had TB or were exposed to TB? Even years ago?
Have you ever known anyone who was treated for tuberculosis? Consumption? Had to be in a sanatorium?
Have you had a TB skin test before? Why? If not for employment, was it part of a TB exposure investigation?
If a TB skin test was done before, was medicine ever prescribed?
Some people remember a relative who had a nurse give them injections for chest infection for many months—this is commonly true for treatment outside the US.
Some people remember that their grandparents had to be in a TB hospital for a long time.

Interpretation of the TST result is influenced by 1) the patient's pretest probability of exposure to TB (e.g., history of incarceration), 2) the risk of developing TB if the patient has latent TB (e.g., infliximab treatment or underlying AIDS), and 3) the influence of comorbidities or medications on the patient's ability to react to the TST. Thus, someone with AIDS only needs 5 mm of induration for the reaction to be considered positive, but a healthy adult without known TB exposure or medical problems must have 15 mm of induration to meet the definition of positive. This approach reduces the proportion of false positives that may occur in areas such as the southeastern US, where much of the population is exposed to environmental mycobacteria, such as Mycobacteria avium complex. A thorough review of the diagnosis and treatment of LTBI can be found on the Centers for Disease Control and Prevention (CDC) Web page (online at The CDC has not made specific recommendations about the amount of induration required for a positive result in RA patients. Patients with RA are often anergic, and take one or more immunosuppressive agents, such as methotrexate and low doses of corticosteroids. Since the addition of infliximab dramatically increases the risk for reactivation of latent TB, a 5-mm cutoff may be a prudent choice if treatment with an anti-TNF agent is being considered.

Testing for anergy by placing Candida albicans or other “control” skin tests at the same time as the TST is no longer recommended (8). If a person starting an anti-TNF agent has a credible history of TB exposure and is already too immunosuppressed to be able to react to a TST, therapy for LTBI may be a prudent course of action, even if the TST result is negative. For example, known exposure to a member of the household who had pulmonary TB would be cause for concern. However, if patients report having complied with an approved regimen of TB preventive therapy, they may not need further evaluation or treatment.

Once it is determined that a person's TST result is positive, active disease must be ruled out. The person should be screened for symptoms consistent with active disease, and appropriate microbiologic, radiographic, and pathologic studies should be obtained. If the person has symptoms suggestive of TB, especially cough, a simple surgical mask should be placed. If a chest radiograph shows an infiltrate or other suggestion of active TB, the patient should remain in respiratory isolation, and sputum should be collected and sent for AFB smear and culture. The patient's local health department should be notified. If there is a strong suspicion of active TB disease, 4-drug TB therapy should be started while awaiting culture results. If there is not a high suspicion of active TB, but the sputum culture results are pending, preventive therapy should be postponed until active TB has been ruled out to avoid causing the development of resistance by the TB bacteria.

When active disease has been ruled out by history, radiography, and, in some cases, sputum culture, the risk for progression to disease again must be assessed, to decide whether the risk of preventive therapy outweighs the risk of progression to active TB. If the person is to be treated long-term with the equivalent of prednisone at a dosage of ≥15 mg/day or with methotrexate, then treatment of LTBI (also called preventive therapy) should be strongly considered. If treatment with a TNF inhibitor is being considered, then treatment of LTBI would be strongly recommended. Furthermore, since the greatest number of cases of TB have been associated with infliximab, it would be best if this agent could be avoided. It is also likely that combining corticosteroids, methotrexate, and anticytokine agents has an additive effect, increasing the likelihood of progression to active TB.

The bulk of the data supports treatment for 9 months with isoniazid (INH), which decreases the rate of developing TB by at least 70%; 12 months of INH does not appear to improve the efficacy significantly, and 6 months of INH is effective, although the effectiveness is probably closer to 60% (9). A 2-month regimen of RIF plus pyrazinamide (PZA) has been tested in human immunodeficiency virus (HIV)–infected populations and found to be similar in efficacy to 12 months of INH (10). This regimen was accepted and recommended by the CDC as an alternative regimen independent of HIV status, but reports of unexpected hepatotoxicity caused the CDC to retreat from its initial full endorsement of this regimen. Finally, the efficacy and safety of RIF alone for 4 months have not been tested in large, placebo-controlled trials, but RIF has been used successfully in a number of circumstances (11, 12), with an estimated efficacy rate of 63%.

In most population cohorts, hepatotoxicity appears to occur most frequently with the combination of RIF and PZA, followed by a 10% likelihood of hepatitis with INH. Underlying alcoholism, chronic alcoholic or viral hepatitis, and increased age are all associated with an increased risk of hepatitis, which can be severe and fatal. It is possible that an element of immunosuppression, as is seen in HIV/AIDS, may confer some protection against liver toxicity, but this is speculation (13).

A question frequently asked relates to the timing of initiation of a drug such as infliximab once latent TB has been diagnosed and active TB disease has been ruled out. As yet, there are no data from clinical trials to answer the question. Ideally, latent TB will have been diagnosed early in the rheumatologic evaluation process, and preventive therapy will be well under way before intensive immunosuppression is needed. If at all possible, a minimum of 1 month of INH prior to anti-TNF therapy seems prudent, so that drug tolerability and toxicities can be evaluated independently of the anti-TNF treatment regimen. Once a course of preventive therapy has been completed, experience with HIV/AIDS and latent TB supports the recommendation that prolonging preventive therapy is unnecessary (for review, see ref.9), even if lengthy and intense immunosuppression is likely.

Another question is whether it is wise to even consider treatment with anti-TNF agents in the setting of LTBI. Thus far, the greatest risk for TB disease has been seen with infliximab, and very few cases have been reported with etanercept (Enbrel: Immunex, Seattle, WA). The early data with Humira therapy indicate that its associated TB risk will fall between those of the two older anti-TNF drugs, leading to more TB cases than with Enbrel and fewer than with Remicade. There is another anticytokine option, anakinra (Kineret; Amgen, Thousand Oaks, CA), which is an anti–IL-1 receptor agent with no associated cases of TB reported thus far. In most cases, the severity of the RA will dictate which agent is chosen, but consideration of LTBI should be part of the decision-making process. No matter which agent is chosen, if a clinician recognizes the presence of latent TB, initiates preventive therapy, and watches the patient carefully for signs of TB disease, the risks, even with infliximab, may be acceptable and manageable in individual circumstances.

Clinical diagnosis of TB

The cytokines elaborated during infection are probably responsible for many of the symptoms associated with clinical TB. Typical symptoms of TB include fatigue, weakness, and anorexia, and signs include weight loss and fever that is often low grade, more commonly causing night sweats than rigors. Pulmonary inflammation produces an increasingly productive cough and sometimes an associated pleural effusion that may cause chest pain. However, in patients receiving anticytokine therapy, or in those whose cell-mediated immunity is no longer intact because of AIDS or severe malnutrition, for example, the signs and symptoms may be more subtle. Fever may be masked, and cough may be limited or nonproductive. Patients usually feel quite unwell, however, and a clinical picture of “failure to thrive” in a patient receiving infliximab should generate a thorough evaluation for infection. The clinician should assess the onset of illness in relationship to the initiation of anti-TNF treatment. In the report by Keane et al (7), infliximab-associated cases of TB were recognized at a median of 12 weeks after the first infusion, and a median of 3 infliximab infusions were received prior to TB diagnosis.

In the setting of immunosuppression, the body's inability to form a protective nodule around the tubercle bacillus (granuloma) leads to unabated growth of the organism and its dissemination to a variety of organs. Disseminated tuberculosis causes progressive, severe disease, often with unusual manifestations, which may lead to a delay in diagnosis, and death. In the setting of AIDS and TB, it is common to be able to grow the organism from the blood if AFB blood cultures are requested. It is too early to know whether mycobacteremia will be common in patients treated with anti-TNF agents, but sending a blood culture for AFB testing is reasonable when evaluating a fever of unknown origin in this setting. Other sites where TB can manifest include the hepatobiliary system, bones (Pott's disease), and central nervous system.

Radiographic investigations should be guided by the patient's symptoms and physical examination findings. Abnormalities should be pursued with an aim toward obtaining tissue for pathologic evaluation and microbiologic culture. If cerebrospinal fluid (CSF) is sampled, at least 10 ml should be sent to the TB laboratory for culture. It is rare to see TB bacteria when staining the CSF, even in cases that are eventually culture-confirmed. The role for nucleic acid amplification (NAA) or polymerase chain reaction (PCR) in analysis of the CSF is unclear, and neither of these techniques has been approved by the FDA for diagnostic use in suspected cases of TB meningitis.

Clearly, the diagnosis of TB depends on several key findings, ideally culminating in confirmation by microbiologic culture. An algorithm for diagnosing TB is shown in Figure 1.

Figure 1.

Tuberculosis (TB) diagnosis algorithm. Diagnosis of TB depends on several key findings, ideally culminating in microbiologic culture confirmation. Deciding whether and when empiric treatment for TB should start depends on the pace and severity of illness and on the likelihood that TB is causing the symptoms. While awaiting confirmation of a single diagnosis, evaluation and sometimes empirical therapy for more than one opportunistic infection may be appropriate. PPD = purified protein derivative; AFB = acid-fast bacilli.

Microbiologic diagnosis of TB

The first question the microbiology laboratory can answer for a clinician is whether bacteria can be seen under a microscope, which can provide a clue as to the type of infection present. The cell wall of all mycobacteria includes mycolic acids, which retain certain organic dyes despite an acid decolorization step. This property separates mycobacteria from other bacteria, and all mycobacteria are “acid-fast” when stained with a Ziehl-Neelsen stain. They are often referred to as acid-fast bacilli (AFB). Although unique to mycobacteria, this characteristic does not differentiate Mtb from other less-pathogenic or nonpathogenic mycobacteria. Once the staining process has identified the presence of mycobacteria, the organism must be grown in culture to confirm the identity of the species. Furthermore, only about half of the cases of culture-proven TB will have enough mycobacteria present in samples sent to the laboratory to be detected by the staining process. Even if not seen on the initial AFB stain, Mtb growing in culture is always considered a pathogen, not a “colonizer.”

The stain and culture techniques used to detect mycobacteria are completely different from the techniques used when routine bacterial cultures are requested, and they will not be performed unless the clinician specifically requests tests for mycobacteria. Usually, if the clinician requests “AFB,” “rule out TB,” or “mycobacteria” smear, culture, and susceptibility, the correct tests will be done. Currently, most US laboratories use a modified Ziehl-Neelsen stain as well as a fluorochrome stain that is viewed under a fluorescence microscope, enhancing the sensitivity of the test many times. If the test is done on site, the staining results may be available by the end of the day. Conversely, culture results are usually not available for 10–21 days, even using a liquid medium growth system. The higher the burden of organisms, the faster the culture results are known. Once the culture becomes positive, a probe is used to confirm or rule out the presence of Mtb. Drug susceptibility testing requires an additional 10 days or more.

NAA or PCR-type assays are approved for direct use on sputum specimens in the US, and they have good test characteristics when enough organisms are present to be seen by routine AFB staining, although they are less sensitive when fewer bacteria are present (14). Although these amplification assays may be helpful for obtaining an earlier diagnosis, they do not provide susceptibility data, which are only available by growing the mycobacteria in culture. A PCR test should never be ordered without ordering an accompanying AFB smear, culture, and susceptibility test.

Once TB is suspected, it is critical to obtain a tissue specimen to send specifically for TB testing. If the infection is in the lung, sputum can usually be obtained. Induced sputum, using saline or other expectorants, is also valuable. If neither is possible, bronchoalveolar lavage is a useful option; the postbronchoscopy sputum over the ensuing 24 hours is quite valuable and should be collected for AFB smear and culture. If TB is found in extrapulmonary sites, the sputum still should be obtained if possible, to determine the person's infectiousness to others. If the site of investigation requires invasive diagnostic methods, several milliliters of material or pus, or one or more biopsy specimens, must be kept out of formalin and sent specifically for TB culture and testing.

Treatment of TB

Once a case of active TB is suspected or microbiologically confirmed, it must be reported to the local health department's TB program. If a TB expert has not already been consulted, one should be consulted at this point if possible. Telephone advice can be obtained from state and county TB programs or from other reference centers listed on the CDC Web site (online at

Treatment should follow the current guidelines recently published jointly by the CDC, the American Thoracic Society, and the Infectious Diseases Society of America and should include INH, RIF, PZA, and ethambutol (EMB) (15). Inclusion of INH, RIF, and PZA in the initial phase of treatment is essential to be able to complete therapy in 6 months, the goal of short-course TB chemotherapy. All therapy should be given under the direction of a member of the TB treatment team; this approach is called directly observed therapy (DOT). Studies show that short-course DOT is the most successful regimen and that it is recommended throughout the world. In the US, the TB control programs in 48 of the 50 states provide TB medications and monitoring free of charge for TB patients. Daily therapy is given for at least the first 14 days and may then be switched to a twice-weekly or thrice-weekly regimen, with appropriate increase in dosing, to facilitate DOT. Communication between the patient's physician and the public health team is crucial. Once the susceptibility results confirm a fully susceptible isolate, EMB can be discontinued, and 8 weeks of treatment with INH, RIF, and PZA can be completed. After the 8-week intensive phase, PZA should be discontinued, and INH and RIF should be given for an additional 18 weeks.

The public health team should collect sputum every 1 or 2 weeks until culture-conversion has been documented. A person with cavitary TB whose sputum culture takes longer than 8 weeks to convert to negative should receive extended therapy (16). A person with drug intolerance or drug toxicity, or who has drug-resistant TB, must be managed in concert with an expert.

In a severely immunocompromised person in whom TB is suspected, treatment should be initiated while awaiting final culture results. A delay in empirical treatment can cause severe morbidity and excess mortality. If the person is receiving infliximab or other anti-TNF drugs, these agents should be discontinued, at least temporarily. All other immunosuppressive agents should be tapered to the lowest possible dosage. Once TB treatment has been initiated and the patient is responding—usually in 6–8 weeks—the least immunosuppressive antiinflammatory regimen can be reintroduced. There are no data to guide clinicians about post–TB treatment issues in patients who continue to need agents such as infliximab to control severe inflammatory diseases. At this point, there are no recommendations to use prolonged treatment regimens or to use INH alone after TB therapy has been completed.

In conclusion, natural and iatrogenic immunosuppression can markedly alter the likelihood that a person's LTBI will progress to active TB. Clinicians must think of TB as a possibility and work to confirm or eliminate the diagnosis when appropriate. Initiating prompt empirical treatment can be lifesaving. Diagnosing and treating latent TB should be aggressively pursued in patients about to be treated with immunosuppressive agents, especially TNF inhibitors.