Methods

We analyzed culture-positive cases of Mycobacterium tuberculosis infection reported to the California TB registry from 1993 through 2006. XDR TB cases had at least 1 isolate that met the definition of extensively drug resistant [1]. Pre–XDR TB isolates were multidrug resistant (resistant to isoniazid and rifampin) and resistant to either a fluoroquinolone or an injectable agent, but not both. XDR TB evaluable isolates were multidrug resistant, with susceptibilities reported for at least 1 fluoroquinolone and 1 injectable agent.

The Centers for Disease Control and Prevention collects drug-susceptibility results on initial isolates at diagnosis and on the final isolates collected at least 30 days later and compiles these results in a report of verified cases of TB. State law (Health and Safety Code Title 17 §2505) requires MDR TB isolate submission to local public health laboratories and to the California Department of Public Health Microbial Diseases Laboratory for second-line drug susceptibility testing. Selected isolates are sent to the Centers for Disease Control and Prevention for additional drug susceptibility testing. We reviewed all available susceptibility results for TB cases from 1993 through 2006 from local, state, and national laboratories. Patients were included if XDR TB was identified on at least 1 isolate (reported on the Centers for Disease Control and Prevention case report form or on interim laboratory results available to the state TB registry). Drug resistance absent from the initial specimen but found on an interim or final isolate after TB treatment initiation indicated resistance acquired during TB treatment.

To determine whether patients with XDR TB had received prior TB treatment in California, we queried the California TB registry. To determine AIDS status, we matched each XDR TB case with the California AIDS case registry. To find cases that resulted from XDR TB transmission in California, we searched for any person with XDR TB case who had been identified as a contact to another person with XDR TB. Strain genotyping was incomplete for cases before 2000 and was therefore not included in our analysis.

Categorical data were compared using Χ² tests or, if expected cell sizes were small, Fisher's exact test. The Wilcoxon rank sum test was used to determine differences in distribution of continuous variables. The Cochran-Armitage test was used to identify trends in proportions of pre–XDR TB cases from 1993 through 2005. Drug susceptibility testing of final isolates for 2006 was incomplete at the time of this analysis and was excluded from statistical testing of the trend. A P value ⩾.05 was considered to be statistically significant. Univariate analyses were performed with SAS software, version 9.1 (SAS Institute).

Results

Magnitude and trends. Among 52,104 TB cases reported in California from 1993 through 2006 (figure 1), 39,952 (77%) were culture positive for M. tuberculosis complex. Of these, 38,767 (97%) were tested for isoniazid and rifampin susceptibility. Among 627 multidrug-resistant cases, 424 (68%) had susceptibilities reported for at least 1 fluoroquinolone and 1 injectable agent. In this XDR TB evaluable group, 18 cases (4.2%) were extensively drug resistant.

From 1993 through 2006, 0–3 cases of XDR TB were reported each year (figure 2) from 7 of California's 61 local health jurisdictions. The number of pre–XDR TB cases identified was 77, or 18% of all XDR TB evaluable cases (figure 1). The proportion of pre–XDR TB cases among XDR TB evaluable cases increased from 7.1% in 1993 to 32.1% in 2005 (P=.02) (figure 2).

Resistance to second-line drugs. The number of drug susceptibility results (from tests performed on unique isolates or on identical isolates tested at different laboratories) per XDR TB case ranged from 1 through 12, with an average of 4. Initial M. tuberculosis isolates were tested for susceptibility to both a fluoroquinolone and an injectable agent in 17 of 18 XDR TB cases. Twelve (71%) of 17 patients had XDR TB at the time of diagnosis, and 5 (29%) acquired XDR TB during therapy in California. Among the patients who had XDR TB on presentation, 10 (83%) were foreign born. Initial drug resistance patterns for patients with acquired XDR TB included 3 with multidrug resistance and 2 with isoniazid and ethambutol resistance. At diagnosis, many XDR TB cases were resistant to most second-line drugs (table 1). Isolates from 1 case had discordant second-line drug susceptibilities that prevented definitive classification as XDR TB. However, this case was included in our series because it met the surveillance definition of having at least 1 isolate that was extensively drug resistant.

During the study period, second-line susceptibility testing increased. The percentage of multidrug-resistant isolates tested for susceptibility to fluoroquinolones (ciprofloxacin and ofloxacin) increased from 50% in 1993 to 80% in 2004. Similarly, the percentage of multidrug-resistant isolates tested for injectable susceptibility increased from 58% in 1993 to 92% in 2004. The proportion of multidrug-resistant isolates that were evaluated for both a fluoroquinolone and injectable (i.e., evaluable for XDR TB) increased from 45% in 1993 to 71% in 2006 (P⩾.001). However, the proportion of multidrug-resistant isolates with resistance to either a fluoroquinolone or injectable did not change significantly over time. From 1993 through 2004, ciprofloxacin or ofloxacin resistance was reported in 8.5% (range, 6%-12%), and injectable resistance was reported in 14% (range, 6.5%-18%) of all multidrug-resistant isolates.

Demographic characteristics of patients. Patients with XDR TB were largely foreign born (83.3%) and recent immigrants to the United States (table 2). Age ranged from 1 to 79 years (median age, 42 years). Nine patients (50%) were Hispanic, 7 (38.9%) were Asian, 1 (5.6%) was black, and 1 (5.6%) was white. The country of origin was most commonly Mexico (46.7%), followed by South Korea (20%), The Philippines (13.3%), and Vietnam, China, and India, with 1 case each (6.7%). Among 14 XDR TB cases occurring in foreign-born patients with known arrival dates in the United States (93.3%), 6 cases (42.9%) were reported within 6 months after patient arrival to the United States. The median time in the United States at TB report date was 10.8 months. Nine (50%) of the patients with XDR TB had a prior history of TB and were foreign born, with TB occurring 1 to 16 years before arrival in the United States. No patients with XDR TB were identified as having AIDS. Two patients (1 born in the United States and 1 born elsewhere ) were household contacts of another patient with XDR TB, suggesting transmission of XDR TB in California. Three patients with XDR TB (21.4%) were born in the United States: an infant who contracted XDR TB from an immigrant relative, a homeless individual who acquired XDR TB while undergoing treatment, and an incarcerated patient receiving TB treatment before 1993 in California.

Clinical characteristics. All patients had pulmonary disease. Twelve patients (66.7%) had positive smear results, and 6 (33.3%) had cavitation apparent on a chest radiograph (table 2). Sputum culture conversion was documented in 7 (46.7%) of 15 patients, with a median time to conversion of 195 days (table 3).

Outcomes for patients with XDR TB were poor: 5 (29.4%) died, 7 (41.2%) completed therapy, and 4 (23.5%) moved and have unknown outcomes. Destinations of those who moved were Mexico (2 patients), The Philippines (1 patient), and Korea (1 patient). One patient whose condition was diagnosed in 2006 continues to receive therapy and was excluded from the outcomes analysis.

Comparison of patients who have XDR TB with those who have MDR TB. Slight differences were found between patients with MDR TB whose isolates were evaluable for extensive drug resistance and those whose isolates were not tested for XDR TB. Patients with XDR TB–evaluable isolates were more likely to be younger (mean age, 40 vs. 46 years; P=.002), to be recent immigrants (3.2 vs. 5.1 years from US arrival to TB diagnosis; P=.007), to have positive smear results (67.5% vs. 59.1%; P=.04), and to have received directly observed therapy throughout treatment (60.9% vs. 47.9%; P=.005).

Patients in California with XDR TB and those with MDR TB were similar in terms of race, age, and proportions who were foreign born, had positive smear results, or had cavitary lung disease (table 2). However, patients with XDR TB had their conditions diagnosed sooner after arrival in the United States (median, 0.9 year), compared with those with MDR TB (median, 3.7 years) and other TB (median, 9 years) (table 2). Patients with XDR TB were also more likely to have a prior TB diagnosis and to have been born in Mexico than were those with MDR TB and other TB (table 2). Fewer patients with XDR TB had documented culture conversion (46.7%), compared with those with MDR TB (87.3%) and other TB (79.5%) (table 3). Time to culture result conversion for patients with XDR TB who achieved negative sputum culture results was protracted (median, 195 days for patients with XDR TB vs. 98.5 days for those with MDR TB and 60 days for those with other TB (table 3). Finally, more patients with XDR TB died (29.4%) or moved (23.5%) and fewer completed therapy (41.2%), compared with those with MDR TB and other TB (table 3).

Comparison of trends in XDR TB for 1993–1999 and 2000–2006. Although XDR TB occurred primarily in foreign-born individuals throughout the study period, the country of origin differed between patients identified in early and later years. Before 2000, 33% of XDR TB cases occurred in Mexican-born individuals, whereas 67% occurred in immigrants from Asia. Since 2000, 67% of XDR TB cases occurred in immigrants from Mexico, whereas 33% occurred in patients from Asia.

Several notable differences were found in the characteristics of patients with XDR TB between the periods 1993–1999 and 2000–2006. Of the 11 patients with XDR TB identified during 1993–1999, 5 (45%) received all of their treatment via directly observed therapy, 3 (27%) had documented sputum culture result conversion, 3 (27%) completed therapy, and 5 (45%) died. In contrast, among the 7 patients with XDR TB identified since 2000, 6 (86%) received all of their treatment by directly observed therapy (including 1 patient who currently continues to receive treatment), sputum culture result conversion occurred in 4 (57%), 4 (57%) completed therapy, and no deaths were reported.

Discussion

Although XDR TB is a new term, cases have been occurring for >1 decade [3] and are now recognized worldwide [5]. California, with nearly 3000 cases of TB annually, has reported the largest number of MDR TB cases in the nation since 2002 [11]. In this report, we document the extent of XDR TB in California in, to our knowledge, the first statewide descriptive study of XDR TB in the United States. We identified 18 XDR TB cases in California during 1993–2006.

Patients from California with XDR TB had poor outcomes, as seen elsewhere. Of all patients with XDR TB, 29% died; among the 12 patients with known final outcomes, 41% died. Of note, none of the patients in California with XDR TB had AIDS, in contrast to the national [6] and global epidemiology of XDR TB [5]. Lack of HIV coinfection may explain the absence of extrapulmonary disease and longer survival times seen in our series. Although XDR TB was rapidly and uniformly fatal in HIV-infected patients in South Africa [8], the HIV-negative patients with XDR TB in California have lower mortality rates but a protracted infectious period.

XDR TB appears to have occurred in California through importation, acquired drug resistance, and transmission. Eighty-three percent of cases occurred in foreign-born individuals from 6 different countries, and nearly one-half of the patients were recent immigrants who had been in the United States for ⩾6 months at the time of diagnosis, suggesting their disease was acquired in their country of origin. Five patients acquired XDR TB during treatment in California. Although surveillance data provide limited details on factors related to acquired drug resistance, our findings suggest that, even under US program conditions, failure(s) in current TB control practices can create further drug resistance. Finally, transmission of XDR TB within California resulted in 2 cases. Transmission of XDR TB in the United States is a concern when one considers that most patients with XDR TB were smear positive, one-third had cavitary disease, and, on average, each patient with TB has 6–10 close contacts [12].

Important differences exist among patients with XDR TB identified before and after 2000. The proportion with conversion of sputum culture results increased and deaths decreased among patients with XDR TB identified after 2000. Improved outcomes may be attributable to the increase in directly observed therapy after 2000 and experience gained in MDR TB treatment. Although the predominance of foreign-born patients among those with XDR TB remained similar throughout the study period, since 2000, most cases in California involved Mexican-born immigrants. This finding is in contrast to XDR TB cases that occurred before 2000 (these occurred largely among Asians) and differs from recent national data that indicate that most XDR TB cases were reported in Asian patients [6]. The increase in Mexican-born patients with XDR TB in the second half of the study period is probably associated with multiple factors and not simply increased immigration, because the proportion of Mexican-born patients with TB in California remained relatively constant. During each year of our study period, roughly one-quarter of all TB cases and one-third of all TB cases that occurred in foreign-born patients involved Mexican-born individuals. Drug resistance rates in Mexico may affect the frequency of XDR TB in California and the United States and deserve further study.

We also identified an important subset of patients with pre–XDR TB. These patients comprise 18% of XDR TB evaluable multidrug-resistant cases in California and are 1 drug away from becoming XDR TB. Nationwide, the number of pre–XDR TB cases may be large, although such surveillance data have not been published. In a South African study using a convenience sample, 14% of multidrug-resistant isolates were found to be pre–XDR TB [13]. This finding suggests that we must implement strategies to identify and cure patients with pre–XDR TB before they develop XDR TB. Modeling studies support the notion that unless evolution of MDR TB into XDR TB is slowed, the number of cases of XDR TB could increase exponentially [14]. Prevention is also more cost-effective than treatment, given that the cost of inpatient treatment for a single XDR TB case in California is estimated to be $600,000 in 2006 US dollars (A. Nitta; personal communication, based on a survey of 6 local cases).

In addition, we report that testing of MDR TB isolates for fluoroquinolone and aminoglycoside susceptibility increased over time, but the proportion of isolates with resistance to either class of drug did not change significantly. Our finding that 8.5% of multidrug-resistant isolates were fluoroquinolone resistant is higher than has been previously reported in the United States but within the range found in other nations [15, 16]. It is surprising that fluoroquinolone resistance rates did not change significantly, despite increasingly widespread use of these drugs [17]. Similar to our data, some countries report no change in fluoroquinolone resistance rates [16], but others note significant increases in fluoroquinolone resistance among MDR TB isolates [15, 18, 19]. These discrepancies may be due to differences in the survey populations or lack of standardized methods across laboratories for second-line drug susceptibility testing.

Our study has several limitations. The number of XDR TB cases is likely to be an underestimation; because of incomplete second-line drug susceptibility reporting, some XDR TB cases were probably not identified by our surveillance system. Also, nearly one-quarter of patients with XDR TB moved, so we lack data on final outcomes. In addition, HIV-positive status may have been underreported, because we matched XDR TB cases with an AIDS registry, a method that identifies HIV-positive patients but not patients with an unknown HIV status. Finally, our case definition does not account for discordance in second-line drug susceptibilities and may result in misclassification. The XDR TB status of 1 patient in our series was not definitive because of conflicting susceptibility results.

Our description of XDR TB in California has important implications for TB control efforts. The sizeable proportion (32%) of patients with MDR TB who were not tested for both a fluoroquinolone and an injectable agent raises concern that patients are receiving suboptimal treatment, which is likely to contribute to poor patient outcomes and acquisition of further drug resistance. The absence of susceptibility testing may have contributed to the development of XDR TB strains in South Africa [3]. Development of routine, standardized, and rapid methods for drug susceptibility testing and guidelines for reporting and resolving discordant susceptibility results across laboratories are needed. Strengthening cornerstones of TB care, including consistent use of directly observed therapy, treatment by experienced TB control physicians, isolation of infectious patients with XDR TB, and effective contact investigation are also essential for XDR TB control. Follow-up of patients with XDR TB who have moved also deserves further attention, because a large proportion of patients with XDR TB in California moved and had unknown outcomes. There is currently no mandatory state-to-state tracking of patients with TB, although initiatives such as TBNet (http://www.migrantclinician.org/network/tbnet) and CureTB (http://www.curetb.org) are in place to promote continuity across states and between the United States and Mexico.

In California, ongoing efforts to control multidrug-resistant and XDR TB include improving rates of directly observed therapy, using molecular assays for rapid detection of isoniazid and rifampin resistance [20], enhanced surveillance of all patients with XDR TB since 2000 (including medical record review and quality assurance of drug susceptibility reporting), and implementation of a MDR TB consultation service for health care providers. Increased support of domestic and international TB control programs and drug development efforts are urgently needed to combat XDR TB.