Tuberculosis caused by RD^Rio Mycobacterium tuberculosis is not associated with differential clinical features

Abstract

BACKGROUND

We recently described the Mycobacterium tuberculosis RD^Rio genotype, a clonally derived sublineage within the Latin American–Mediterranean (LAM) family. Genetic diversity of M. tuberculosis likely affects the clinical aspects of tuberculosis (TB). Prospective studies that address this issue are scarce and remain controversial.

OBJECTIVE

To determine the association of differential clinical features of pulmonary TB with the RD^Rio M. tuberculosis etiology.

METHODS

Culture-proven pulmonary TB patients (n = 272) were clinically evaluated, including history, physical examination, chest X-ray and anti-human immunodeficiency virus serology. Isolates were classified as RD^Rio or non-RD^Rio M. tuberculosis by multiplex polymerase chain reaction and further spoligotyped. Clinical and M. tuberculosis genotype data were analyzed.

RESULTS

RD^Rio M. tuberculosis caused disease in 26.5% (72/270) of all TB cases. The LAM genotype, of which RD^Rio strains are members, was responsible for 46.0% of the TB cases. Demographic data, major signs and symptoms, radiographic presentation, microbiological features and clinical outcomes were not significantly different among patients with TB caused by RD^Rio and non-RD^Rio strains.

CONCLUSIONS

Disease caused by M. tuberculosis RD^Rio strains was not clinically distinctive or more severe than disease caused by non-RD^Rio strains in this series of TB patients. Larger prospective studies specifically designed to disclose differential clinical characteristics of TB caused by specific M. tuberculosis lineages are needed.

A leading cause of death worldwide, tuberculosis (TB) is primarily a disease of the lungs, caused by very closely related mycobacteria belonging to the Mycobacterium tuberculosis complex (MTC). Genotyping of MTC isolates is used to identify mycobacteria at species level and distinguish MTC into distinct lineages and/or sublineages that are a powerful tool for TB epidemiological control, phylogenetic exploration and investigation of host-pathogen interactions.

The highly structured population genetics of MTC is defined by six major phylogeographic lineages: M. africanum (West-African lineages 1 and 2), M. bovis and four M. tuberculosis lineages: Indo-Oceanic (formerly referred as ancestral), Euro-American, East Asian (mostly Beijing) and East-African Indian (EAI).¹ These major lineages can be definitively characterized using long sequencing polymorphisms (LSPs).² More often, however, for the purposes of TB surveillance, the less precise spoligotyping and mycobacterial interspersed repetitive units-variable number tandem repeats (MIRU-VNTR) are used.

The Latin American–Mediterranean (LAM) genotype is a member of the Euro-American lineage of M. tuberculosis and is the dominant genotype responsible for approximately half of Brazilian TB cases.^3–6 In previous studies, we described that the clonally derived RD^Rio strains of M. tuberculosis are defined by a ~26.3 kb deletion resulting in the loss or modification of 10 genes, including two preproenkephalin genes known to be recognized by host immunity.⁷ In prior publications from our group, insertion sequence (IS) 6110 restriction fragment length polymorphism and MIRU-VNTR typing showed that the RD^Rio sublineage likely occurred at a more distant evolutionary time, as both methods indicated considerable divergence, while it remains in the Euro-American M. tuberculosis lineage as well as belonging to the single nucleotide polymorphism Cluster VI grouping.^6–8

RD^Rio M. tuberculosis strains, present in at least 18 other countries of Europe, Africa and the Americas, are all members of the LAM genotype and a sublineage of the EuroAmerican lineage.^6,7,9 RD^Rio M. tuberculosis is the most prevalent cause of TB in Rio de Janeiro and has also been reported in large numbers throughout other Brazilian regions.^4,7,10 Results from recent studies suggest that M. tuberculosis lineages are adapted to particular human populations, and that M. tuberculosis is older and more genetically diverse than previously assumed.^1,11 It is likely that such genetic diversity could result in lineage-specific biological characteristics, ultimately influencing both the clinical and epidemiological aspects of TB. However, information on the nature of coupling between M. tuberculosis lineages and the development of distinctive clinical manifestations of TB is scarce, and this issue remains somewhat enigmatic. In the present prospective study, we sought to determine whether there were distinct demographic, clinical, radiologic and laboratory features associated with pulmonary TB caused by RD^Rio M. tuberculosis strains.

MATERIALS AND METHODS

Study setting and design

A total of 278 consenting adult patients with culture-proven pulmonary TB, diagnosed at the Federal University of Rio de Janeiro (FURJ) and two community health centers were enrolled in the study. A thorough clinical evaluation, including history, physical examination, chest X-ray (CXR), anti-human immunodeficiency virus (HIV) serology, tuberculin skin test (TST) and patient follow-up for at least 1 year after completion of treatment were performed as part of routine clinical care. CXRs, performed before starting treatment, were independently evaluated and scored in a standardized form by two experienced pulmonologists blinded to the RD^Rio genotype. In cases of disagreement, discussion among readers led to a decision by consensus on the final scoring. Of 217 films, 11 were excluded due to poor quality, and CXRs were not available for analysis in 57 of the 272 TB cases.

The study protocol was approved by the Institutional Review Board of the FURJ University Hospital. Signed informed consent was obtained from all patients.

Phenotypic, biochemical and genetic characterization of M. tuberculosis isolates

Mycobacterial (Löwenstein-Jensen) culture and phenotypical and biochemical characterization of isolates from all patients were performed according to standard methods.¹² Drug susceptibility testing (DST) against anti-tuberculosis drugs was performed using the proportion method.¹³ The average time until M. tuberculosis growth detection in cultures and bacillary load determination using quantitative acid-fast bacilli (AFB) smears and colony-forming unit assays were recorded on admission and at 8 weeks of treatment. A multiplex polymerase chain reaction (PCR) assay of all 278 M. tuberculosis culture thermolysates was performed on these isolates to differentiate RD^Rio from other lineage strains, as described by Gibson et al.⁹ For further genotyping, DNA from M. tuberculosis isolates was extracted and spoligotyping performed using PCR and hybridization conditions on a commercially available membrane (Isogen Bioscience BV, Maarssen, The Netherlands), as previously described.¹⁴ Spoligotype patterns were recorded both in a 43-digit binary format representing the 43 spacers and as an octal code.¹⁵ M. tuberculosis patterns were introduced into a Microsoft Office Excel datasheet (Microsoft, Redwoods, WA, USA) for cluster recognition and classification and then compared with the SpolDB4 database of the Pasteur Institute of Guadeloupe (http://www.pasteur-guadeloupe.fr:8081/SITVITDemo) for type classification.

Data analysis

Data were summarized by median and standard deviation for continuous variables and by frequency and proportion for categorical variables. Categorical variables were compared for RD^Rio and non-RD^Rio sublineages using the χ² and Fisher’s exact tests. The Mann-Whitney test was applied for continuous variables. Odds ratios (ORs) and their 95% confidence intervals and P values were estimated. Statistical analysis was performed using GraphPad Prism, version 4.02 (GraphPad Software, La Jolla, CA, USA). P ≤ 0.05 was considered significant.

RESULTS

Pulmonary M. tuberculosis isolates were obtained from sputum (72.1%), induced sputum (25.4%) and bronchoalveolar lavage fluid (2.5%). All examined isolates, corresponding to the 278 enrolled culture-positive pulmonary TB patients, were characterized as M. tuberculosis using biochemical methods, further submitted to multiplex PCR and grouped as RD^Rio or non-RD^Rio sublineage. Six patients with pulmonary specimens presenting two different strains and/or an indeterminate RD^Rio/non-RD^Rio genotype were excluded from the final analysis. The RD^Rio M. tuberculosis was the cause of disease in 26.5% of the patients in this series. Spoligopatterns were generated for 189/272 isolates (70.2%); all presented genotypes characteristic of M. tuberculosis. Altogether, the LAM family was responsible for 46.0% of the TB cases in this series. Based on spoligotype analysis, all RD^Rio strains belong to LAM 9, LAM 1 and/or LAM 2 strains, whereas the LAM 3 genotype was found exclusively among non-RD^Rio strains. In addition to LAM, the T, Haarlem, U, X, Beijing and Beijing-like families were observed among non-RD^Rio strains (Table 1). No orphan patterns were observed.

Table 1.

M. tuberculosis spoligotype families and their frequencies among non-RD^Rio and RD^Rio TB pulmonary TB patients

Spoligotype	Non-RDRio-positive (n = 139) n (%)	RDRio-positive (n = 50) n (%)
LAM (Type 1321)	0	1 (2.0)
LAM 1	0	7 (14.0)
LAM 2	0	5 (10.0)
LAM 3	12 (8.6)	0
LAM 3 S/convergent	2 (1.4)	0
LAM 4	5 (3.6)	5 (10.0)
LAM 6	13 (9.4)	0
LAM 9	5 (3.6)	32 (64.0)
Haarlem 1	11 (8.0)	0
Haarlem 3	34 (24.6)	0
Haarlem 4	2 (1.4)	0
U	6 (4.3)	0
U (likely H)	1 (0.7)	0
X1	2 (1.4)	0
X2	2 (1.4)	0
X3	2 (1.4)	0
S	2 (1.4)	0
Beijing	1 (0.7)	0
Beijing Like	1 (0.7)	0
T4-CEU1	3 (2.2)	0
T5_MAD2	1 (0.7)	0
T3	3 (2.2)	0
T2–T3	2 (1.4)	0
T1	29 (20.9)	0

TB = tuberculosis; LAM = Latin American–Mediterranean.

Overall, clinical data showed that most patients were AFB smear-positive (79.4%) and had advanced disease on CXR, with infiltrates involving more than one lobe (66.5%) and/or cavitation(s) (53.8%). The rate of TB-HIV co-infection was 8.5%. The characteristics of patients infected with RD^Rio M. tuberculosis strains (n = 72) were compared with those infected with non-RD^Rio strains (n = 200). Socio-demographic and clinico-radiological data are summarized in Table 2. The majority of our patients were male (68.0%). Age, sex, HIV status, CXR appearance (including presence of cavitations and disease extension) and frequencies of smoking, bacille Calmette-Guérin scar, positive TST, response to treatment, previous anti-tuberculosis treatment and/or recurrence after completion of a previous standard TB treatment were not significantly different among patients with TB caused by RD^Rio and those with non-RD^Rio strains. On entering the study, patients did not differ significantly as regards presence of cough, chest pain, expectoration, hemoptysis, weight loss, fever (Table 2), or duration of symptoms (data not shown). Interestingly, anorexia was significantly more frequent among RD^Rio TB patients; however, the nutritional status in both groups of patients was similar, as judged by body mass index (BMI; 19.9 vs. 21.3 kg/m²) and weight gain during the first 2 months of treatment. Microbiological features of RD^Rio and non-RD^Rio pulmonary TB patients are presented in Table 3. The average time until M. tuberculosis growth detection in cultures, bacillary load and frequencies of positive AFB smears, single-drug resistance and multidrug-resistant TB (MDR-TB) were very similar among both groups of TB patients.

Table 2.

Socio-demographic and clinico-radiological features of non-RD^Rio and RD^Rio pulmonary tuberculosis patients at the time of diagnosis

	Non-RDRio		RDRio
	Total n	Positive n (%)	Total n	Positive n (%)	P value	OR (95%CI)
Cases, n	200	200 (100)	72	72 (100)
Demographic features
Female sex	200	70 (35)	72	17 (23.6)	0.08	0.57 (0.31–1.1)
Age
Median [range]	200	41 [18–86]	72	39.0 [19–72]	0.46	—
Median, males	130	44.0	55	39.0	0.01	—
Median, females	70	34.0	17	42.0	0.08	—
Clinical features
Signs and symptoms at the time of diagnosis
Cough	195	164 (84.1)	72	58 (80.5)	0.49	0.78 (0.39–1.6)
Sputum	188	129 (68.6)	64	43 (67.2)	0.83	0.94 (0.51–1.7)
Hemoptysis	186	31 (16.7)	65	13 (20.0)	0.54	1.30 (0.61–2.6)
Chest pain	162	80 (49.4)	56	23 (41.1)	0.28	0.71 (0.39–1.3)
Fever	191	112 (58.6)	67	41 (61.2)	0.71	1.1 (0.63–2.0)
Sweats	175	91 (52.0)	57	27 (47.4)	0.54	0.83 (0.46–1.5)
Anorexia	149	76 (51.0)	49	15 (30.6)	0.01	0.42 (0.21–0.84)
Weight loss	184	145 (78.8)	67	55 (82.0)	0.57	1.2 (0.60–2.5)
CXR features at the time of diagnosis*	157		49
Extension of pulmonary involvement
Normal CXR		2 (1.3)		0	1.0	0.63 (0.03–13.3)
≤1 pulmonary lobe		46 (29.3)		21 (42.8)	0.08	1.8 (0.93–3.5)
>1 pulmonary lobe		109 (69.4)		28 (57.1)	0.11	0.59 (0.30–1.1)
Cavitary disease	157	85 (54.1)	49	26 (53.0)	0.89	0.96 (0.50–1.8)
Previous anti-tuberculosis treatment status
Never treated	194	156 (80.4)	69	59 (85.5)	0.35	1.4 (0.67–3.1)
Relapse after previous treatment (<5 years)	194	24 (12.4)	69	8 (11.6)	0.86	0.93 (0.40–2.2)
Body mass index on admission, median	148	19.9	57	20.3	0.50	—
BCG scar	175	115 (65.7)	64	42 (65.6)	0.98	1.0 (0.54–1.80)
TST (>5 mm)	144	108 (75.0)	50	39 (78.0)	0.67	1.2 (0.55–2.5)
Current smoker	195	46 (23.6)	71	23 (32.4)	0.15	1.6 (0.85–2.8)
HIV co-infection	153	14 (9.1)	45	3 (6.7)	0.77	0.71 (0.19–2.6)
Diabetes mellitus	190	26 (13.7)	67	9 (13.4)	0.96	0.98 (0.43–2.2)

^*Of 217 films, 11 were excluded due to poor quality reasons and CXR results were not available for analysis in 57.

OR = odds ratio; CI = confidence interval; CXR = chest X-ray; BCG = bacille Calmette-Guérin; TST = tuberculin skin test; HIV = human immunodeficiency virus.

Table 3.

Microbiologic features of non-RD^Rio and RD^Rio pulmonary tuberculosis patients

	Non-RDRio		RDRio
	Total n	Positive n (%)	Total n	Positive n (%)	P value	OR (95%CI)
AFB smear on admission
Spontaneous sputum	140	127 (90.7)	56	50 (89.3)	0.76	0.85 (0.31–2.37)
Induced sputum or BAL	60	34 (57.6)	16	5 (31.25)	0.09	0.33 (0.10–1.08)
Bacillary load on admission*	196†		72
1–19 colonies		19 (9.7)		10 (13.9)	0.33	1.5 (0.66–3.4)
1+		77 (39.3)		23 (31.9)	0.27	0.73 (0.41–1.3)
2+		52 (26.5)		14 (19.4)	0.23	0.67 (0.34–1.3)
3+		48 (24.5)		25 (34.7)	0.09	1.6 (0.91–2.9)
Median time to growth detection, days	200	(21.0)	72	(21.5)	0.58	—
Culture-positive at 2 months of treatment	57	14 (24.6)	20	5 (25.0)	1.0	1.02 (0.32–3.33)
Drug resistance‡
RMP	193	7 (3.6)	70	1 (1.4)	0.69	0.39 (0.05–3.2)
INH	191	20 (10.5)	70	3 (4.3)	0.14	0.38 (0.11–1.3)
Ethambutol	188	5 (2.7)	69	1 (1.4)	1.0	0.54 (0.06–4.70)
Streptomycin	193	17 (8.8)	70	6 (8.6)	0.95	0.97 (0.37–2.6)
Ethionamide	176	3 (1.7)	61	0	0.57	0.40 (0.02–7.9)
MDR-TB§	170	6 (0.3)	61	1 (0.04)	0.46	0.46 (0.05–3.86)

^*1+ = 20–100 colonies; 2+ = 100–200 colonies; 3+ ≥201 colonies.

^†4 cultures performed using Mycobacteria Growth Indicator Tube 960 were excluded.

^‡Pyrazinamide susceptibility testing was not performed.

^§Resistance to both INH and RMP.

OR = odds ratio; CI = confidence interval; AFB = acid-fast bacilli; BAL = bronchoalveolar lavage; RMP = rifampin; INH = isoniazid; MDR-TB = multidrug-resistant tuberculosis.

DISCUSSION

M. tuberculosis has evolved in large, divergent clonal lineages that are geographically constrained. Importantly, specific ethnic groups seem to have distinct susceptibilities to particular M. tuberculosis strains.^1,6,11,15 Thus, along with interactions of social, environmental and host genetic backgrounds, the M. tuberculosis genetic diversity likely affects both the clinical and epidemiological aspects of TB in a lineage-specific way. The M. tuberculosis Euro-American superlineage, in which the LAM genotype family is strongly predominant, is the leading etiology of TB in Brazil. In a recent study, it was identified in respectively 96.4% and 66.8% of the analyzed isolates.⁴ We recently described that the RD^Rio genotype, defined by an ~26.3 kb deletion, is a clonally derived sublineage of the Euro-American superlineage and a member of the LAM family.^6,7,9 The M. tuberculosis RD^Rio sublineage is the most prevalent cause of TB in Rio de Janeiro and is also present in other Brazilian and world regions.^6,8,9,16,17 In accordance with our previous results,^6,7,9 the present work confirmed RD^Rio exclusively as a LAM sublineage. Indeed, RD^Rio comprised more than half of the LAM strains (57.4%). Most of the strains with an LAM9 spoligopattern were of the RD^Rio sublineage (86.5%), and all LAM1 and LAM2 spoligotypes were RD^Rio sublineage, while the LAM3 spoligotype was exclusively of the non-RD^Rio genotype.

In the present study, pulmonary TB caused by the M. tuberculosis RD^Rio sublineage in Rio de Janeiro was not clinically distinctive or more severe than TB caused by non-RD^Rio M. tuberculosis strains with respect to patient demographic data, major signs and symptoms, radiographic presentation, microbiological features, frequency of resistant strains and treatment outcome. Although lineage-specific differences in host ethnic preference and TB epidemiology are beginning to emerge,^{1,6,11,18–21} studies correlating M. tuberculosis-specific lineage to differential clinical patterns of TB are scarce and often present conflicting results. Differential radiographic patterns have been reported by some,^10,22,23 but not others.^6,17,18 It was observed that Indonesian TB patients infected with Beijing strains more often had a febrile response to treatment,¹⁸ while the contrary was reported in Russia.²³ In Taiwan, no difference in treatment outcomes were found between W-Beijing and non-W-Beijing groups.¹⁹ In fact, with the exception of a few studies that have looked at the M. tuberculosis lineage etiology of extra-pulmonary TB,^20,22 frequencies of MDR-TB^21,23,24 and progression to severe disease,^11,16,22 most data have shown preponderantly marginal effects on few clinical parameters.^7,10,17–19 In addition, the drug resistance observed in studies looking at MDR-TB and lineages may be acquired and not clonally derived.²⁵

Overall, our results are in agreement with previously published works that failed to disclose a clear-cut and/or specific association between M. tuberculosis lineage and major TB clinical features.^{7,10,17–19,26,27} Nevertheless, for several reasons, these results do not convincingly preclude the importance of M. tuberculosis lineage specificity in TB pathogenesis: 1) there is evidence from in vitro^28,29 and in vivo^30–32 research suggesting that genotype plays a role in pathogenicity; 2) epidemiological studies indicate that some M. tuberculosis genotypes, including RD^Rio, may have different transmission rates favoring unequal distribution of M. tuberculosis;^{6,11,18,22,24} 3) we do not know that all strains belonging to a particular lineage will share the same pathogenicity factors; 4) genetically distinct specific M. tuberculosis lineage/sublineage strains may be circulating; 5) some associations seem to be true to some patient populations but not others;^10,18,23 6) socio-economic and cultural barriers to effective medical care may increase the time elapsed before diagnosis, masking major clinical differences; 7) often only a small number of patients are analyzed thoroughly and in depth; 8) longitudinal specific study design is rarely used; and 9) most studies lack data on a key variable: TB recurrence after completion of adequate anti-tuberculosis treatment in a context of high treatment completion rates.

In the present study, response to current treatment, default rate, previous anti-tuberculosis treatment and/or recurrence after completion of a previous standard anti-tuberculosis regimen were not significantly different among patients with TB caused by RD^Rio and those with non-RD^Rio M. tuberculosis strains. In addition, microbiologic parameters, such as the average time until detection of M. tuberculosis growth in cultures, bacillary load in cultures and frequencies of positive AFB smears before starting treatment and after 2 months of treatment, and frequencies of single or multiple drug resistance, were similar. Although criticized, 2-month sputum culture is the most widely used putative surrogate for prediction of treatment outcome and relapse.³³ The similar bacillary load of sputum cultures of RD^Rio and non-RD^Rio TB patients at 2 months of treatment suggests that, considering this somewhat limited criterion, a major, clinically relevant difference in virulence between RD^Rio and non-RD^Rio strains is unlikely. This is in agreement with in vitro fitness analysis comparing these strains.²⁸

The frequencies of medical conditions that have been associated with increased TB risk, namely HIV infection, diabetes mellitus and tobacco smoking, were similar in both groups, suggesting that RD^Rio strains were not more infective and/or virulent among patients presenting one of these causes of immune dysfunction. Among all clinical parameters analyzed, only anorexia showed a significantly different frequency among the two groups, occurring much more frequently among non-RD^Rio TB patients. However, the nutritional status in both groups was similar, as based on BMI, and the rate of weight gain during the first 2 months of treatment was also comparable.

The lack of sublineage association with most major clinical parameters (MDR-TB, HIV co-infection, clear clinical and microbiological differences in response to treatment), observed in the present work, are in agreement with most of our earlier results on RD^Rio TB.^6,7,10 However, the previously described association of lung cavitary disease,^7,10 hemoptysis and higher bacillary load with RD^Rio TB were not observed in the present study. Possible explanations include sample size, differences between this study and prior studies in the ratio of non-RD^Rio LAM and other genotypes in relations to RD^Rio strains, nutrition, and social changes occurring in Rio de Janeiro and Brazil in general.

This study was designed with the following power (80–90%, α = 0.5%) and sample size considerations, but was terminated because the anticipated grant did not materialize. The sample size estimated for hemoptysis (RD^Rio 30% vs. non-RD^Rio 10%) was 250–313, for cavitary disease (RD^Rio 40% vs. non-RD^Rio 25%) it was 627–783, and for sputum bacillary load 3+ (RD^Rio 20% vs. non-RD^Rio 10%) it was 627–783. Of note, in the current study, sputum bacillary load 3+ ORs for RD^Rio was 1.6 (P = 0.09), despite the fact that this study did not attain the anticipated sample size. In addition, in our prior studies in Brazil, the proportion of RD^Rio genotype strains was 39% and 37%,^6,16 while in this study, the proportion of RD^Rio genotype strains was 31%. Importantly, the total cohort for our first study in Rio de Janeiro had 310 unique patient isolates, while this study had 272. Lower sample size and/or the lower proportion of isolates with the RD^Rio genotype could therefore have affected our ability to demonstrate differences in this study. The present study finding that anorexia was associated with harboring non-RD^Rio genotype M. tuberculosis may indicate that the current study also had the above limitations, or it may be due to the better nutritional status of the Brazilian population owing to the improved socio-economic conditions in Brazil in the last 10 years.

CONCLUSIONS

Pulmonary TB caused by M. tuberculosis RD^Rio sublineage in Rio de Janeiro was not clinically distinctive or more severe than TB caused by non-RD^Rio M. tuberculosis strains in this series of patients. However, this study had several potential limitations that may have contributed to the lack of differences being observed. Larger prospective studies specifically designed to disclose differential clinical characteristics of TB caused by specific M. tuberculosis lineages are thus needed. Such studies would allow conclusive or at least stronger causal inferences regarding the clinical specificity of M. tuberculosis lineages/sublineages.