INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

After many years of focusing on Mycobacterium avium complex (MAC) lung disease that mimicked tuberculosis (i.e., upper lobe cavitary disease in the male with a history of alcohol and heavy cigarette abuse), recent attention has been on MAC lung disease as it appears in patients with no apparent risk factors. Beginning with the study by Prince and coworkers, which appeared in 1989 (1), much has been learned about the epidemiology and radiographic features of this latter disease. The patients are primarily elderly women with no history of smoking, and reticulonodular infiltrates that predominantly involve the right middle lobe and lingula (1). By high-resolution computed tomography (HRCT) these patients have patchy bilateral bronchiectasis, with routine involvement of the right middle lobe and lingula (3, 5, 7).

Whether the bronchiectasis is the risk factor for MAC or is a consequence of bronchial involvement by the disease is the source of ongoing debate.

While doing studies of pulsed field gel electrophoresis (PFGE) of pretreatment and relapse isolates in patients participating in macrolide treatment trials for MAC, we noted that some patients had more than one PFGE pattern. We then decided to formally study the incidence of single or multiple infecting strains in MAC lung disease in human immunodeficiency virus (HIV)-seronegative patients, focusing on the nodular interstitial form of MAC lung disease.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Patients were chosen from participants in one of several ongoing macrolide treatment trials of HIV-seronegative patients with M. avium complex lung disease at the University of Texas Health Center. Details of these studies are published elsewhere (8). The clinical trials have been ongoing since December 1991, and approximately 250 patients have been enrolled since that time. A recent study of acid-fast bacilli (AFB) sputum smears suggests that approximately 50% of cases in our population have nodular bronchiectasis and approximately 50% have upper lobe cavitary disease (11). As part of these trials, detailed histories and lung radiographs were obtained pretreatment and as treatment progressed. Patients were chosen for the current study who had typical radiographic features of the upper lobe (apical) cavitary form or nodular bronchiectasis form, of M. avium complex lung disease (3, 5, 7); who had at least five pretreatment and/or on treatment sputum isolates of M. avium complex available for study; and for whom dates and records of the sputum laboratory results were available. The presence of other potential or opportunistic pathogens sputum specimens was also recorded. Initially, equal numbers of patients in each group were planned. With the surprising results of the nodular bronchiectasis group, it was expanded to an approximate 2:1 ratio with the cavitary group.

Organisms

The design of the treatment trials was that three sputum cultures were to be obtained before drug therapy, at approximately monthly intervals while on therapy, and as needed after drug therapy (8). Some of these isolates were then saved by freezing at 70° C. Between 5 to 10 isolates were chosen for each patient from the frozen stocks, attempting to spread them over as long a period of time as possible.

Sputum samples were processed according to standard methods of digestion and culture. Staining for AFB was by the fluorochrome method, and samples were cultured on Middlebrook 7H10 agar and in BACTEC 12B broth. Primary cultures on Middlebrook 7H10 agar or 7H10 agar subcultures of BACTEC 12 B bottles were saved by swabbing with a cotton swab across the agar surface (hereafter referred to as sweep cultures), subculturing once to be sure no contamination was present, and freezing the organisms at 70° C in trypticase soy broth with 15% glycerol at the time of isolation.

PFGE

Isolates were taken from the frozen stocks, subcultured to agar, then sweep cultures were used for the studies. Patterns of large restriction fragments of genomic DNA were obtained by PFGE as described previously (12, 13). Briefly, organisms collected from liquid cultures were cast into low-melting-point agarose plugs, then lysed by lysozyme (1 mg/ml) sodium dodecyl sulfate (1%), and proteinase K (1 mg/ml). Genomic DNA contained in the plugs was digested with the restriction endonucleases DraI and XbaI and separated by PFGE with CHEF Mapper system (Bio-Rad Laboratories, Richmond, CA) at 14° C for 20 h at 6 V/cm. Selected strains were also digested with AseI and/or SpeI. Pulse time was ramped from 3 to 12 s after XbaI digestion, from 5 to 15 s for 14 h and then from 60 to 70 s for 6 h after DraI digestion, from 5 to 20 s after AseI digestion, and from 3 to 12 s after SpeI digestion. The gel was stained with ethidium bromide, then photographed.

For patterns that appeared to reflect the presence of more than one genotype, 10 single colonies were isolated and their clones subjected to PFGE. Single-colony isolations were also made from several cultures that did not appear mixed by PFGE as a control.

The definitions of PFGE strain relatedness were those of Tenover and coworkers (14). With a minimum of 10 interpretable bands, strains were interpreted as indistinguishable (no band differences), closely related (1-3 band differences), possibly related (4-6 band differences), and different (seven or more band differences). Individual PFGE types identified by PFGE were referred to as genotypes.

M. avium Complex and Species Identification

Organisms were initially identified as M. avium complex, using a commercial DNA probe (Accuprobe; GenProbe, Inc., San Diego, CA). They were included for study only if they were positive with the M. avium complex group probe. Each M. avium complex genotype was then typed as M. avium, M. intracellulare or X group (MAC-X), initially using the two species-specific commercial probes (Accuprobe). Confirmation of the species was made using a previously described multiplex polymerase chain reaction (PCR) (15). Each genotype was also evaluated using the PCR-restriction fragment length polymorphism patterns (PRA) of a 439 bp fragment of the 65 kD heat shock protein gene as described by Telenti and coworkers (16). Genotypes were considered to belong to MAC-X if they were positive with the M. avium complex probe but negative with the specific M. avium and M. intracellulare probes, and if they were then negative for M. avium and M. intracellulare using the multiplex PCR (15).

Serotyping

Cultures shown to consist of a single PFGE pattern (genotype) were subcultured from the frozen stocks, then sweep cultures subjected to serotyping. Serotyping was performed by a microtube agglutination technique using antisera representing M. avium (serovars 1-6, 8-11, and 21), M. intracellulare (serovars 7, 12-20, 22-28), and M. scrofulaceum (serovars 41- 43) (17). Genotypes that spontaneously agglutinated with multiple antisera were characterized as rough, and those that failed to agglutinate with a recognized antisera were termed nontypable.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Seventeen patients were chosen with the nodular bronchiectatic form of M. avium complex lung disease. Clinical details of the patients are shown in Table 1. The patients were predominantly female (76%), had predominant right middle lobe and/or lingular interstitial/nodular infiltrates (100%), had patchy bilateral cylindrical bronchiectasis on HRCT (12 of 12 patients tested), and were elderly (13 of 17 or 76% were over 60 yr of age at the time of diagnosis). Only five patients (29%) had ever smoked with only one (6%) having smoked more than 20 pack-yr. No patients abused alcohol.

Most patients had between 50-100 sputum specimens submitted for culture. Eight of the 17 patients (47%) with nodular bronchiectasis had other nontuberculous mycobacteria or other potential pathogens associated with bronchiectasis recovered on two or more occasions during the course of laboratory evaluation. Five patients had M. abscessus, two patients had M. kansasii, two patients had Nocardia spp., two patients had Pseudomonas aeruginosa, and one patient each had M. fortuitum and Aspergillus sp.

Nine patients were chosen who had the cavitary form of M. avium complex lung disease. Clinical details of these patients are shown in Table 2. The patients were predominantly male (78%), and all (100%) had large upper lobe cavitary disease with disease diagnosis before the age of 60. All but one patient (89%) were smokers, and had smoked more than 20 pack-yr. Most patients (78%) abused alcohol.

A similar number of sputum specimens were submitted for culture as with the nodular bronchiectasis group. No patient with upper lobe cavitary disease had other potential pathogens including nontuberculous mycobacteria recovered on two or more occasions during the course of their respiratory illness.

Organisms

A total of 124 sputum cultures from the 17 patients with nodular bronchiectasis were studied (a mean of 7.3 cultures per patient). Cultures were obtained before, during, and after completion of drug therapy over a range of 8.5 mo to 49 mo (mean 26.1 mo). The specimens from which the isolates were recovered generally contained small numbers of organisms, with only 17% being AFB smear positive (fluorochrome method) and 44% being recovered only from BACTEC 12 B bottles and/or resulted in less than 10 colonies on 7H10 agar (see Table 3).

A total of 69 M. avium complex cultures were studied from the nine patients with cavitary disease, a mean of 7.7 isolates per patient. Isolates were obtained before, during, and after therapy over a range of 9 mo to 45 mo (mean of 22 mo). Unlike the specimens from patients with nodular bronchiectasis, the specimens from patients with cavitary disease which served as the source of the cultures generally contained large numbers of organisms, with 70% being AFB smear positive and only 9% having fewer than 10 colonies on 7H10 agar (see Table 3).

An additional 88 single-colony clones from 8 cultures from seven patients with nodular bronchiectasis were studied (18 clones from one patient, 20 clones from another patient, 10 per culture for the remainder). Six cultures were studied which appeared to have a mixed PFGE pattern, while two cultures which gave a single PFGE pattern were chosen as controls.

Pulsed Field Gel Electrophoresis

A total of 49 genotypes were recovered from the 17 patients with nodular bronchiectasis, a mean of 2.9 genotypes per patient. The range was 1 to 9 genotypes, with 15 of 17 (88%) of patients having two or more genotypes and 9 of 17 (53%) having 3 or more genotypes. Multiple genotypes in the same sputum culture were detected in 3 of 17 (18%) of the patients (an example is seen in Figure 1). Multiple genotypes in the same patient were identified in specimens before or during therapy for 13 of 15 (87%) of the patients with multiple genotypes, and after sputum conversion and discontinuation of drug therapy with 2 of 15 (13%) of the patients with multiple genotypes. An example of multiple genotypes from the same patient is shown in Figure 2. The recovery of multiple genotypes over time is shown for another patient in Figure 3.

View larger version (129K): [in this window] [in a new window]   Figure 1.   Example of multiple PFGE types (genotypes) in the same sputum specimen. DraI PFGE patterns of six sputum cultures from a patient with nodular bronchiectasis over a period of 3 yr and 10 mo. Pattern a is present in lanes 2 and 3, pattern b is present in lane 6, while the cultures in lanes 1, 4, and 5 are a mixture of patterns a and b; lane 7, DNA standards.

View larger version (87K): [in this window] [in a new window]   View larger version (86K): [in this window] [in a new window]   Figure 2.   Example of multiple PFGE patterns (genotypes) in different sputum specimens from the same patient. DraI (A) and XbaI (B) PFGE patterns of nine genotypes recovered from a single patient with nodular bronchiectasis. Lanes 1 to 9, genotypes recovered, received in a time period of 3 yr and 3 mo from a single patient; lane 10, DNA standards.

View larger version (13K): [in this window] [in a new window]   Figure 3.   PFGE patterns over time in a 69-yr-old female patient with nodular bronchiectasis who had three genotypes of M. avium complex in her initial cultures (patterns a, b, and c), converted her sputa to negative on treatment, then relapsed off therapy with one of the original patterns (b).

A total of 11 genotypes were recovered from the nine patients with cavitary disease, a mean of 1.2 genotypes per patient. The results from a typical patient with a single PFGE pattern are shown in Figure 4. The timing of PFGE patterns in another patient is shown in Figure 5. Only 1 of 9 patients (11%) had multiple genotypes. One of the three genotypes in this patient was recovered after successful drug therapy on one occasion. It was AFB smear negative and grew only in the BACTEC 12 B bottle.

View larger version (106K): [in this window] [in a new window]   Figure 4.   Example of single PFGE pattern (genotype) typically seen in patients with upper lobe cavitary disease. XbaI PFGE patterns of six isolates from a single patient with cavitary disease received in a period of 14 mo. Lanes 1 to 6, patient isolates; lane 7, control isolate from a different patient; lane 8, DNA standards.

View larger version (11K): [in this window] [in a new window]   Figure 5.   PFGE patterns over time in a 58-yr-old male with upper lobe cavitary disease due to M. avium complex who remained positive despite 3 yr of treatment (Rx).

The large restriction fragment (LRF) bands were compared for all isolates for each patient considered by the Tenover (14) criteria to have either indistinguishable or related strains (and have given the same strain designation). For the 17 patients with nodular bronchiectasis, 13 had indistinguishable isolates (no band differences) with DraI, and XbaI. Three patients had 1-3 band differences with either enzyme for some isolates (closely related). One patient had 0-2 LRF band differences with DraI and 2-6 band differences with XbaI (closely related and possibly related). For the eight patients with cavitary disease, seven had indistinguishable isolates (no band differences) with both enzymes and one patient had a 1 band difference with DraI only for 3 of 10 isolates.

Of the six cultures that appeared to have a mixed pattern consisting of more than one genotype present on the same PFGE gel, five were shown to contain multiple genotypes when 10 to 18 single colony clones were studied. Three cultures consisted of 2 genotypes, while two cultures consisted of 3 genotypes. Two cultures chosen as controls contained only a single genotype when 10 single-colony clones were studied.

Species Identification

All 60 genotypes studied were positive with the DNA probe for M. avium complex. Of the 49 genotypes identified among patients with nodular bronchiectasis, 47 were identified to species (Table 4). Thirty-four genotypes (72%) were M. intracellulare, 10 genotypes (21%) were M. avium, and 3 genotypes (6%) were MAC-X. Mycobacterium intracellulare was the predominant genotype (i.e., the genotype from each patient identified in the greatest number of cultures) from 15 of 17 patients (88%).

The findings among patients with cavitary disease were almost identical. Of the 11 genotypes identified among patients with cavitary disease, 8 (73%) were M. intracellulare, 2 (18%) were M. avium, and 1 (9%) was MAC-X. Mycobacterium intracellulare was the predominant genotype in 8 of 9 (89%) of the patients.

Complete agreement was evident among the DNA probes, the multiplex PCR, PRA for all isolates identified as M. avium or M. intracellulare. Agreement was also present with the DNA probes and multiplex PCR for the 4 genotypes of MAC-X. No recognized PRA pattern has yet been determined for this group with PRA.

Serotyping

Of the 49 genotypes identified among patients with nodular bronchiectasis, 42 were subjected to serotyping (see Table 1). Of the 30 genotypes of M. intracellulare, 16 (53%) were either nontypable or rough. There was no predominant serovar among M. intracellulare that were serotypable, with only serovars 7, 19, and 22 appearing more than once. Serovars that were encountered were 7, 12, 13, 16, 19, 22, 24, and 42. Of the 9 genotypes of M. avium, 8 (89%) were serotypable with 5 of 8 being serovar 8.

Of the patients with cavitary disease, 4 of the 7 genotypes of M. intracellulare and 2 of 2 genotypes of M. avium were rough or nontypable, results that were comparable to patients with nodular bronchiectasis.

A surprising finding was that four patients had different genotypes (by the definition of Tenover and coworkers [14]) that were of the same serovar: one patient had two genotypes of serovar 8 (M. avium) (Patient 6, Table 4); one patient had three genotypes of M. intracellulare that were rough (Patient 7); one patient had two genotypes of M. intracellulare serovar 19 (Patient 11); and one patient had two genotypes of M. intracellulare serovar 22 (Patient 12). Review of the PFGE and serotyping data demonstrated these findings to be correct. A summary of the large restriction fragment band differences for the genotypes of these four patients with DraI and XbaI is shown in Table 5. A comparison of the PFGE patterns for three of the patients is shown in Figure 6. The three rough strains from Patient 8 shared few common bands and appeared unrelated (different). The other three pairs of isolates differed by 9 to 16 bands with either DraI or XbaI but shared 11 to 20 bands in common, suggesting they may be related. The most unusual were the two isolates of serovar 22 from Patient 12 which were identical except for the presence of 11 extra bands in one of the strains, suggesting a large deletion from one of the strains. Thus identical serovars with different genotypes (by current definitions) (14) was seen with 4 of 16 (25%) of patients in both disease groups with multiple genotypes that had been subjected to serotyping.

View larger version (109K): [in this window] [in a new window]   View larger version (112K): [in this window] [in a new window]   Figure 6.   PFGE patterns of isolates from the same patient of the same serovar but different PFGE patterns. The chromosomal DNA was digested with DraI (A) and XbaI (B). Lanes 1 and 2, isolates of serovar 22; lanes 3 and 4, isolates of serovar 19; lanes 5 and 6, isolates of serovar 8; lane 7, DNA standards.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies of disseminated M. avium in HIV-seropositive patients have demonstrated that polymicrobial infection occurs in this setting. In 1993 Arbeit and coworkers (18) reported PFGE results on 28 cultures from blood, stool, and sputum from 14 patients, of whom 13 were bacteremic. Three individual colonies from the primary specimen were saved and studied by PFGE. Two of 13 blood cultures and 1 of 6 stool cultures contained more than one strain, providing evidence of polymicrobial infection in 3 of 14 (21%) of patients (18). In an update published in 1994 (19), these investigators had identified polymicrobial infection in 9 (24%) of 37 patients. In a study of clarithromycin resistance among patients with disseminated M. avium, studies of individual colony clones and sweep cultures revealed polymicrobial infection among blood culture isolates in 3 of 16 (19%) of patients (13). More recently in 1997, Mazurek and coworkers (20) reported results from 85 isolates of blood, stool, and respiratory secretions from 25 patients with acquired immunodeficiency syndrome (AIDS) from San Francisco. Comparison of two single-colony clones with a sweep culture revealed mixed cultures in only 2 of 27 blood specimens, 4 of 8 respiratory specimens, and zero of six stool specimens. Overall, 8 of 25 (40%) patients had multiple strains (none had more than two) when all isolates from all sources were compared. Mazurek and coworkers were skeptical of the significance of the second strain (genotype) in these patients as they were rarely recovered from more than one specimen. In a study of 196 blood isolates with M. avium complex from 93 AIDS patients, Picardeau and coworkers identified polymicrobial disease in only three patients (3.2%) using multiple techniques including PFGE and IS 1245 hybridization (21). They comment that the level of bacteremia was low, so only a small number of individual colonies could be studied. A mean of 3.7 cultures per patient over a mean of 6.8 mo were compared.

Because PFGE is labor-intensive, the number of cultures and individual colonies (clones) studied for their LRF pattern in these studies has been limited, and the incidence of mixed infections is probably higher. In an autopsy study of AIDS patients who died with active disseminated M. avium disease, Torriani and coworkers cultured blood and multiple organs from five patients. With PFGE examination of 85 single-colony clones, 3 of the 5 (60%) had two strains in different organs (22), supporting this belief.

This is the first study to do genetic comparisons of multiple isolates of M. avium complex from patients with chronic lung disease and to relate it to the type of clinical disease. It clearly demonstrates that patients with the nodular bronchiectasis form of the disease are infected with multiple strains of M. avium complex. Given the multifocal nature of the disease within the lung and the apparent association of the nodular disease with areas of bronchiectasis, the potential for infection of different sites of the lung with different strains seems quite plausible. Whether these multiple strains are present in different sites or the same site, however, has not been studied.

In contrast, 8 of 9 (89%) of patients with cavitary disease were infected with a single strain of M. avium complex. The one patient with multiple genotypes of M. avium complex had a second genotype recovered once on therapy, and a third genotype recovered from a single BACTEC bottle after resection of her cavity and after long-term sputum conversion from the initial infection. Radiographic disease in these patients was limited to the upper lobes, and most had a predominantly single confluent area of fibrocavitary disease. In a previous study by Mazurek and coworkers (12), seven isolates collected over a 23-mo period from a patient with cavitary disease showed the presence of only a single strain by PFGE.

For the eight patients with nodular bronchiectasis in whom multiple genotypes were recovered from sputum cultures with at least two of the genotypes being recovered on more than one occasion on different days (Table 4), the likelihood that one of the genotypes is a laboratory or transient environmental contaminant is reasonably excluded. For the four patients with 3 or more genotypes of which only one was recovered more than once, the likelihood that all of these were contaminants also seems highly unlikely. The greatest concern about contamination would be in the three remaining patients whose second strain was recovered from only a single culture. These single positives were all AFB smear negative, and two of the three grew more than 30 colonies on the solid media as well as growing in the BACTEC 12B broth. Ongoing studies of the significance of positive cultures from M. avium complex in sputum using both Middlebrook 7H10 agar and BACTEC 12B broth suggest that single positive sputum cultures for M. avium complex in patients not at apparent risk for M. avium complex disease (and hence presumed to be contaminants) are AFB smear negative and grow only in the 12B broth. The few that do grow on solid agar grow fewer than 10 colonies (unpublished data). Thus two of these three seem likely to have significant growth when compared with these controls. A laboratory mix-up on whose culture was frozen or recultured can also not be excluded, but neither circumstance was identified among the patients with cavitary disease, suggesting that most of the isolates recovered only once were likely present in the patient.

The issue of colonization versus disease for each genotype was not addressed. Colonization implies the presence of the organism but with no damage to the host. Although the absence of tissue invasion has been established clinically and histopathologically for some potential pathogens such as Aspergillus spp., in the setting of bronchiectasis, it has not been established for nontuberculous mycobacteria. And unlike Aspergillus, M. avium complex produces a definite tissue invasion in this setting. For patients in whom the new genotypes are sequential and follow successful drug therapy of the first genotype, the invasive nature of the second genotype is easier to assess. Three of the current patients were culture negative for > 12 mo after chemotherapy before a new genotype of M. avium complex appeared. Those with multiple isolates clearly had disease associated with the new genotype. For simultaneously recovered genotypes, cultures with genotyping and histopathologic studies of lung biopsies may be necessary to resolve this issue. The transient presence of M. avium complex in a damaged airway (i.e., colonization) seems possible, but the presence of the same strain over long periods of time seems likely to be associated with disease.

A previous study (23) of 154 cases of (primarily) M. abscessus chronic lung disease demonstrated that most patients had identical clinical profiles (female, elderly, nonsmokers with probable bronchiectasis) and radiographic features (nodular interstitial disease) as the patients with the nodular bronchiectatic form of M. avium complex lung disease. Interestingly, 15% of all the patients and approximately 20% of patients with M. abscessus also had M. avium complex recovered from sputum (23). Although no published data are available, it is our impression that a comparable percentage of patients with the nodular bronchiectatic form of M. avium complex will also have M. abscessus (unpublished data). (Five of 16 or 31% of the current M. avium complex patients also had multiple positive sputa for M. abscessus.) This suggests that the risk factor for the two diseases is similar, and makes it logical that patients who have pulmonary disease with more than one species of nontuberculous mycobacteria could have pulmonary disease with more than one genotype of the same species. In contrast, we have not seen a patient with the upper lobe fibrocavitary tuberculous form of M. avium complex who is simultaneously infected with M. abscessus (R.J.W., D.E.G., unpublished data) and none were identified in the current study.

A number of implications are apparent from this study. Drug treatment of the nodular bronchiectatic form of M. avium complex lung disease is much more complicated because multiple strains are involved, and not all strains may respond in a like fashion to the drug combination. Relapses after apparent successful drug therapy will need to be genotyped, as an apparent relapse may be reinfection with another strain and hence not a treatment failure. Single positive cultures in patients with nodular bronchiectasis with different PFGE patterns will need careful clinical assessment as they may reflect the presence of a second infected strain, and not just contamination. The risk for reinfection almost certainly is higher with this form of the disease than with the tuberculous type of upper lobe cavitary disease.

Approximately 70% of genotypes from patients with either nodular bronchiectasis or cavitary lung disease in the current study were M. intracellulare. Studies of M. avium complex isolates collected in the 1980s in the United States (24) and several other countries (25, 26) using serotyping by seroagglutination demonstrated that approximately two-thirds of respiratory isolates in non-AIDS patients with lung disease were M. avium, and only one-third were M. intracellulare. Recent studies with the commercial DNA probe (GenProbe) have confirmed the classification of the serotypes as to which belong to M. avium and which belong to M. intracellulare (27, 28). Studies that used the DNA probe continued to suggest M. avium was the major pathogen in this setting both in the United States and worldwide (27, 29, 30). One of these studies (27) which evaluated isolates from multiple geographic areas in Japan demonstrated that most isolates were M. avium, but with significant geographic variation. Isolates in northern and central Japan were primarily M. avium, while those recovered in southern Japan (Shikoku and Kyushu) were predominantly M. intracellulare. Such geographic differences may explain the high incidence of M. intracellulare in the current study as it was done in a single geographic area (northeast Texas).

The finding that four patients with nodular bronchiectasis were infected with M. avium complex of different genotypes but the same serovar was surprising. In three of the four cases a number of bands were shared in common. It is unknown at present if genotypes of the same serovars show greater similarity than genotypes of different serovars, one possible explanation of this finding. Another possible explanation is that the apparently different genotypes began as a single genotype but then underwent genetic changes (insertions, deletions, etc.) to produce the observed differences. These changes could have taken place in the environment prior to infection as well as in the patient after infection. Most PFGE studies of mycobacteria and other species have shown relative stability of the genomic DNA, suggesting that events producing changes in more than three large restriction fragment bands are highly unusual (14). The observation that most epidemiologically unrelated strains of M. avium complex produce highly diverse PFGE patterns suggests that the restriction sites studied are not highly conserved. It is possible that the time frame for detectable changes may be shorter than previously appreciated, at least in some strains.