Elderly patients with tuberculosis in a low-incidence country – Clinical characteristics, inflammation and outcome

Summary

Background

Susceptibility to respiratory infections increases with age. Diagnosing and treating tuberculosis in the elderly comes with the challenges of fewer specific symptoms and possibly more side effects of treatment. Much is unknown when it comes to tuberculosis in the elderly, especially in relation to inflammation, which may impact mortality. We, therefore, investigated a clinical cohort of elderly tuberculosis patients.

Methods

Patients aged ≥65 years, admitted to our tuberculosis reference center between 2005 and 2021, were retrospectively included in our cohort. Sociodemographic data, clinical characteristics, laboratory results, including inflammatory markers at baseline (monocyte, neutrophil, lymphocyte count, and CRP levels), and treatment outcomes were collected. They were compared to the National Dutch TB Registry and analyzed using descriptive statistics. Survival analysis was performed using univariate Cox regression analysis and a log-rank test. Results were visualized in Kaplan-Meier curves.

Results

104 elderly tuberculosis patients, mostly European, with a mean age of 75 years, were included. None were HIV-infected. Miliary tuberculosis cases were overrepresented (14 %) compared to the National Dutch TB Registry (5 % in elderly, 2 % adults). Fever occurred in 77 % (57/74), and the duration of fever decreased with age. Innate immune markers, including monocyte/lymphocyte-ratio, moderately correlated with CRP. Overall mortality was 15 %, and highest (33 %) in patients with CRP levels > 100 mg/mL.

Conclusion

In elderly tuberculosis patients in a low-incidence setting, mortality rates are higher in comparison to younger patients. The overrepresentation of miliary tuberculosis may suggest waning immunity, with a subset of patients exhibiting strong inflammation associated with increased mortality.

Introduction

The growing elderly population is likely to cause a rise in the number of tuberculosis (TB) cases among the elderly.^1,2 Respiratory infections in the elderly, which are defined as individuals aged 65 years or older,³ contribute significantly to mortality.⁴ This is allegedly due to aging of the immune system, along with factors such as frailty, comorbidities, and structural changes in the lungs, which are believed to increase the risk of pulmonary infections and the development of active TB.^5,6

TB diagnosis in the elderly is challenging as classical symptoms occur less frequently.^7–9 In a post-mortem study in Switzerland, 62 % of TB cases among elderly individuals had not been diagnosed during their lifetime.¹⁰ Data on active TB in hospitalized elderly patients are scarce, especially from low TB incidence countries.^9,11–13

We performed an exploratory in-hospital cohort study on clinical characteristics, inflammation, and outcomes in a low TB incidence country. Using the National Dutch TB Registry (NTR) as a reference, we assessed differences with younger TB patients and whether our in-hospital data could be extrapolated to the entire population. Additionally, we explored the correlation among inflammatory markers, and the association with outcome.

Methods

We performed a retrospective cohort study. Elderly TB patients treated in our TB expert center from 2005 to 2021 were included. Data was retrieved from paper records until 2013, and then electronic files were used. Patients with microbiologically confirmed TB, or a high suspicion of TB and positive treatment response within 8 weeks, were included. Collected data included demographics, comorbidities, risk factors, laboratory and microbiological results, side effects, treatment details, duration, and outcomes. This data was compared to elderly (> 65 years) and adult patients (18–65 years) in the NTR. For detailed definitions and statistics, please refer to the Supplementary Methods.

Results

We included 104 elderly TB patients with a median age of 75 years (Table 1). The majority of patients in our cohort (64 %) and the elderly patients from the national TB registry (NTR, 57 %) were of European descent, compared to only 28 % of adult patients in the NTR. Almost all patients in our cohort presented with co-morbidities.

Table 1. Patient characteristics, clinical data and laboratory results.

Patient characteristics	Cohort (> 65 y) (N = 104)	NTR (> 65 y) (N = 2382)		NTR (18–65 y) (N = 12,292)
	N [%]	%	P-value•	%
Years [Median, IQR]	75[70–80]	NR		NR
Male sex	58 [56 %]	58 %	0.649	59 %
Origin
European	64 [63 %]	57 %	0.361	28 %
Asian	21 [20 %]	20 %	0.958	24 %
African	14 [14 %]	18 %	0.235	40 %
North/South American	5 [5 %]	5 %	0.931	7 %
Comorbidities
Diabetes mellitus	35 [34 %]	NR		NR
Chronic lung disease	27 [26 %]	NR		NR
Malignancy	18 [17 %]	NR		NR
Chronic kidney failure	21 [20 %]	NR		NR
(Solid) organ transplantation	3 [3 %]	NR		NR
HIV^	0 [0 %]	NR		NR
Risk factors
Ever smoking	31/62 [50 %]	NR		NR
Alcohol use	9/46 [9 %]	NR		NR
History of tuberculosis	20 [19 %]	5 %	< 0.001	5 %
Immunosuppressive therapy	17 [16 %]	NR		NR
TB site
PTB	48 [46 %]	47 %	0.863	45 %
PTB + EPTB	25 [24 %]	14 %	0.004	12 %
EPTB	31 [29 %]	39 %	0.059	43 %
EPTB locations
Miliary	14 [14 %]	5 %	0.002	2 %
Central nervous system	8 [8 %]	2 %	0.004	2 %
Spinal	7 [7 %]	NR		NR
Lymph node	10 [10 %]	NR		NR
Pleural	4 [4 %]	4 %	0.942	4 %
Other	13 [13 %]	NR		NR
Species
M. tuberculosis	70 [67 %]	63 %	0.370	64 %
M. bovis#	11 [11 %]	2 %	< 0.001	1 %
Culture negative†	17 [16 %]	29 %	0.005	28 %
No culture performed†	6 [6 %]	6 %	0.922	7 %
Susceptibility
Drug sensitive$	76 [73 %]	98 %	< 0.001	94 %
RR-/MDR-TB	4 [4 %]	1 %	< 0.001	2 %
Isoniazid mono-resistance	1 [1 %]	2 %	0.450	5 %
Unknown	23 [22 %]	NR		NR
Treatment regime
HR(Z)E	57 [55 %]	NR		NR
Alternative	47 [45 %]	NR		NR
Drug toxicity
Hepatotoxicity	27 [26 %]	10 %	< 0.001	5 %
Ocular	2 [2 %]	1 %	0.369	1 %
Psychological	8 [8 %]	1 %	< 0.001	1 %
Neurological	8 [8 %]	2 %	< 0.001	0 %
Treatment outcomes*
Treatment success	86 [83 %]	66 %	< 0.001	86 %
Transferred out	2 [2 %]	6 %	0.082	9 %
Death during treatment¥	16 [15 %]	18 %	0.494	1 %

Legend: patients from our cohort, with reference data from adult TB patients within the National Tuberculosis Registry (NTR) from 2005 to 2021, stratified for elderly patients (> 65 years) and adult patients < 65 years.

^{^}Tested in 94/104 patients (90 %).

^#M. bovis (5 patients) and M. bovis BCG (6 patients) were combined

^†In patients where cultures were negative or not performed, diagnosis was made by suggestive histopathology, positive PCR or with a positive IGRA accompanied by a clinical suspicion for tuberculosis infection.

^$Including natural resistance for pyrazinamide in M. bovis patients.

^*In the NTR (> 65 years), 5.2 % of patients discontinued treatment, and in 4.4 %, data on treatment outcome was missing.

^¥One patient died of malignancy, two patients died of pneumonia/sepsis, and the cause of death was unknown in six patients.

^•The p-value shows the statistical differences between our group and the elderly in the NTR.

NTR = Netherlands tuberculosis register, NR = not reported, RR-TB = rifampicin resistant tuberculosis, MDR-TB = multidrug resistant tuberculosis.

Mycobacterium tuberculosis complex cultures were positive in 69 patients (66 %). Five patients (5 %) had Mycobacterium bovis infections, and six patients (6 %) had Mycobacterium bovis-BCG infections following intravesical immunotherapy. Drug resistance occurred in 4 % of our patients and in only 0.4 % of patients in the same age group in the NTR.

TB was confined to the lungs in 46 % (48/104) of our patients, which is comparable to both age groups in the NTR. Miliary TB was present in 14 % (14/104) of our patients, in contrast to 5 % in patients from the same age groups and 2 % in adults in the NTR.

Fever was present in 77 % (57/74) of patients during admission, with a median duration of 9 days and a maximum measured temperature of median 38.5 °C (IQR 37.9–39.2 °C). Median monocyte/lymphocyte-ratio was 0.51 [IQR 0.35–0.79]. Lymphopenia, which was present in 41 % of patients, in 37 % of PTB patients versus 46 % of EPTB patients. With increasing age, fever duration was shorter, and this difference could not be explained by variations in leukocyte ratios or CRP levels (Fig. 1A). However, CRP, as a marker for innate immune activation, did correlate with higher monocyte/lymphocyte-ratios and neutrophil/lymphocyte-ratios, as well as higher maximum temperature levels and longer fever duration (Fig. 1A).

Fig. 1. Inflammatory parameters in elderly TB patients.

(A) Correlation matrix of continuous variables, using pairwise complete Spearman correlation for clinical variables at the start of treatment. Data completeness was 100 % for age, 71 % for temperature, 50 % for leukocyte differentials, and 98 % for baseline CRP. P-values < 0.05 are indicated with an asterisk. (B) Kaplan-Meier curve depicting survival rates in months for patients with CRP levels < 100 mg/mL versus ≥100 mg/mL.

A standard antibiotic treatment regimen was provided to 55 % (57/104) of the patients. Alternative regimens were chosen in cases of side effects (35 %), resistance (5 %), and anticipated toxicity due to pre-existing comorbidities (7 %). In 27 patients (26 %), one of the first-line drugs was replaced with a quinolone. Seven patients (7 %) were treated with RZE, eight patients (8 %) with HRE, and four patients (4 %) with only two of the first-line antibiotics. Patients with rifampicin resistance were treated with a bedaquiline-containing regimen. One patient with severe TB, where quinolone was contraindicated, received additional treatment with amikacin. An increased dose of rifampicin (20–35 mg/kg) was administered to 26 % of the patients (27/104) following our local guidelines for selected cases with hard-to-treat diseases, based on several studies, including those conducted at our institution.^14,15 Hepatotoxicity occurred in 27 patients: 32 % (22/68) of patients on the normal rifampicin regimen versus 19 % (5/27) with an increased dose of rifampicin. In comparison, the NTR showed a 10 % hepatotoxicity rate in the same age group and 5 % in adults. Other common toxicities were neurological and psychological side effects, both occurring in 8 % of patients (8/104). Ocular toxicity occurred in only 2 % of patients (2/104), not because of optical neuropathy.

The median follow-up was 10 months (IQR 6–22), and the median treatment duration was 6 months (IQR 6–9). Treatment was successful in 82 % (85/104) of patients. Two patients were transferred out of our care. Fifteen percent (16/104) of patients died during treatment, of which 38 % (6/16) were directly attributable to TB. The median survival duration was 101 days [IQR 23–137]. In contrast, elderly patients in the NTR show lower treatment success rates (66 %), with a mortality rate of 18 % and 6 % being transferred out, while mortality rates were only 0.7 % in younger patients < 65 years in the NTR.

CRP, for which data was almost complete and which correlated with the inflammatory markers monocyte/lymphocyte-ratio and neutrophil/lymphocyte-ratio, was used to stratify patients using CRP ≥100 mg/mL as a cut-off.¹⁶ Univariate analysis using Cox regression showed CRP > 100, ML-ratio and chronic liver failure as predictors for mortality. Six-month survival in patients with CRP levels ≥100 mg/mL was 67 % (8/12), compared to 92 % (81/88) in those with a lower CRP (log-rank p = 0.004, Fig. 1B).

Discussion

This is one of the few studies to combine clinical data and inflammatory markers in a low-incidence setting among hospitalized elderly TB patients. Our cohort is characterized by a high proportion of miliary TB compared to those in the Dutch National TB Registry (NTR). Mortality is significant amongst our elderly patients, especially so in the subset of patients with a pronounced inflammatory response.

A large proportion of our patients presented with extra-pulmonary (co)localization, similar to the NTR and other low-incidence settings (40–50 %), but higher than in high-incidence settings (21–22 %).^11,13,17,18 Immunity waning with age likely explains the high rate of EPTB disease, including miliary TB. Moreover, waning immunity could also explain the reactivation of previous TB infections per se. As the majority of our patients are of European descent, and TB incidence steeply declined after the 1950s, we presume that most cases represent reactivation.

Elderly TB patients less often exhibit typical symptoms than younger patients (fever, cough, weight loss, and night sweats).^9–11 However, in our cohort, fever was present in the majority of patients in whom data was retrieved, and we observed a shorter duration of fever as age increased. A study on pneumonia patients also reported a shorter fever duration and a 0.15 °C decrease in maximum temperature with each decade of age increase,¹⁷ which we could not confirm as we only included a 3-decade age span.

Elderly-specific TB risk factors include acquired structural pulmonary changes, comorbidities, frailty, and past exposure. The aging immune system causing impaired immunity, also termed “immunosenescence,” may contribute to the likelihood of TB reactivation and could be a possible cause for the higher occurrence of miliary TB in our patients. Simultaneously, the aging immune system is also characterized by “inflammaging,” a chronic low-grade proinflammatory state. Indeed, elevated neutrophil count, immunoglobulin content, and IL-6 levels in the bronchoalveolar lavage fluid were found in healthy elderly individuals compared to younger ones.¹⁹ Also in circulating blood, higher concentrations of IL-6 and IL-1Ra cytokines were found, together with impaired lymphocyte function diminished, indicated by lower production of IFN-γ and IL-22 after pathogen stimulation.⁶ A study from India showed a potential link between diabetes in TB patients and distinct inflammation compared to healthy individuals and TB controls without diabetes.²⁰ Our small sample size hindered exploring diabetes’s impact on inflammation.

Increased pro-inflammatory cytokine and also CRP levels have been found to indicate frailty and predict mortality in healthy elderly.^20–22 In TB patients, CRP levels > 100 mg/L were associated with TB-related mortality in Denmark,¹⁶ a finding that we also observed. Additionally, in Japan, elevated CRP levels predicted in-hospital mortality, even as a continuous variable.²³ In our patients, CRP levels correlated positively with monocyte/lymphocyte-ratio. Increased monocyte/lymphocyte-ratio, were associated with increased age in one setting,²⁴ and were shown to precede active TB^25–27 and to predict tuberculous meningitis mortality.²⁸ While total monocyte counts do not seem to increase with age, non-classical monocytes do²⁹ and further contributing to an increased monocyte/lymphocyte-ratio, is the lymphopenia we observed in EPTB and most outspokenly in miliary TB cases (data not shown). The strong correlation of monocyte/lymphocyte-ratio with CRP, along with the significant result in the univariate analysis, might point to immunosenescence (with lymphopenia as a proxy) and inflammaging (monocytosis) as factors contributing to poor outcomes.

Although we observed more liver toxicity in our patients, possibly partially attributable to more frequent monitoring in our hospitalized setting compared to the NTR, the number of patients finishing treatment was higher and comparable to other cohort studies reporting similar success rates.^8,11,13 In our elderly patients, mortality rates were higher compared to younger patients in the NTR. This observation has also been noted in other cohorts.^6–8

Limitations of our study are the retrospective design precluding more advanced immunological assessment and constraints in comparing our data with the NTR due to the unavailability of matching. Our results from a third-tier referral hospital, may not be generalizable to other centers. In conclusion, our cohort of hospitalized elderly TB patients presented more often with miliary tuberculosis and had higher mortality rates compared to younger patients in the Dutch low-incidence setting. Elevated CRP levels seem to indicate a poorer prognosis. The inflammatory response in elderly TB patients warrants further investigation to improve the outlook for these patients.

Supplementary Material

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jinf.2024.106200.

Table 2. Univariate Cox regression analysis results for predictors of mortality during the first six months of treatment.

Variable	OR [CI]	P-value
Age per year increase	0,975 [0,894-1,063]	0,562
Male sex	2,589 [0,701-9,565]	0,154
European	0,875 [0,278-2,758]	0,820
BMI	0,983 [0,811-1,191]	0,862
Previous TB	1,184 [0,259-5,402]	0,828
Miliary TB	1,301 [0,285-5,941]	0,734
Diabetes	0,950 [0,286-3,155]	0,933
Chronic liver disease	5,051 [1,364-18,700]	0,015
Pulmonary and extrapulmonary TB	0,166 [0,021-1,331]	0,091
Extrapulmonary TB only	0,729 [0,193-2,749]	0,641
Monocyte-lymphocyte ratio (2log ML-ratio)	2,435 [1,107-5,353]	0,027
CRP (>100mg/ml)	5,072[1,483-17,452]	0,010
Albumin (g/L)	0,627 [0,115-3,422]	0,590
Drug sensitive TB	0,846 [0,436-1,638]	0,619
Standard treatment	2,921 [0,791-10,790]	0,108