Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=ilal20

Download by: [Cornell University Library] Date: 10 May 2017, At: 15:16

Leukemia & Lymphoma

ISSN: 1042-8194 (Print) 1029-2403 (Online) Journal homepage: http://www.tandfonline.com/loi/ilal20

Risk and impact of tuberculosis in patients withmultiple myeloma*

Chun-Kuang Tsai, Leh-Kiong Huon, Shuo-Ming Ou, Ai-Seon Kuan, Chiu-MeiYeh, Yu-Ting Lee, Yao-Chung Liu, Tzeng-Ji Chen, Jin-Hwang Liu & Chia-Jen Liu

To cite this article: Chun-Kuang Tsai, Leh-Kiong Huon, Shuo-Ming Ou, Ai-Seon Kuan, Chiu-Mei Yeh, Yu-Ting Lee, Yao-Chung Liu, Tzeng-Ji Chen, Jin-Hwang Liu & Chia-Jen Liu (2017):Risk and impact of tuberculosis in patients with multiple myeloma*, Leukemia & Lymphoma, DOI:10.1080/10428194.2017.1312369

To link to this article: http://dx.doi.org/10.1080/10428194.2017.1312369

View supplementary material

Published online: 09 May 2017.

Submit your article to this journal

View related articles

View Crossmark data

ORIGINAL ARTICLE: CLINICAL

Risk and impact of tuberculosis in patients with multiple myeloma�Chun-Kuang Tsaia , Leh-Kiong Huonb,c, Shuo-Ming Oud,e, Ai-Seon Kuanf, Chiu-Mei Yehg, Yu-Ting Leee,g,Yao-Chung Liue,g, Tzeng-Ji Chene,h, Jin-Hwang Liug,i,k† and Chia-Jen Liug,j†aDepartment of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan; bDepartment of Otolaryngology—Head & Neck Surgery,Cathay General Hospital, Taipei, Taiwan; cSchool of Medicine, Fu Jen Catholic University, Taipei, Taiwan; dDivision of Nephrology,Department of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan; eSchool of Medicine, National Yang-Ming University, Taipei,Taiwan; fCancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Oxford, UK; gDivision ofHematology, Department of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan; hDepartment of Family Medicine, TaipeiVeterans General Hospital, Taipei, Taiwan; iInstitute of Biopharmaceutical Sciences, National Yang-Ming University, Taipei, Taiwan;jInstitute of Public Health, National Yang-Ming University, Taipei, Taiwan; kChong Hin Loon Cancer and Biotherapy Research Institute,National Yang-Ming University, Taipei, Taiwan

ABSTRACTWe investigated the risk and impact of mycobacterium tuberculosis (TB) infection in patientswith multiple myeloma (MM). We identified 3979MM patients from Taiwan’s National HealthInsurance database between 2000 and 2011 and compared the incidence rates of TB infection inthese patients with 15,916 randomly selected age-, sex-, and comorbidity-matched subjects with-out MM. The risk of TB was higher in the myeloma cohort (adjusted hazard ratio [HR] 3.11, 95%confidence interval [CI] 2.41–4.02). Risk factors for MM patients contracting TB were age �65(adjusted HR 1.93, 95% CI 1.19–3.15), alcohol use disorder (adjusted HR 2.86, 95% CI 1.24–6.62),and steroid daily dose equivalent to prednisone 5mg or more (adjusted HR 2.38, 95% CI1.50–3.77). MM patients with TB had a higher mortality risk than those without (adjusted HR2.03, 95% CI 1.54–2.67). The incidence of TB is significantly higher in MM patients.

Abbreviations: aHR: adjusted hazard ratio; COPD: chronic obstructive pulmonary disease; HSCT:hematopoietic stem cell transplantation; HIV: human immunodeficiency virus; LHID2000:Longitudinal Health Insurance Database 2000; MM: multiple myeloma; NHIRD: National HealthInsurance Research Database; NHRI: Taiwan’s National Health Research Institutes; RCIP: Registryfor Catastrophic Illness Patients; TB: tuberculosis

ARTICLE HISTORYReceived 4 October 2016Revised 26 February 2017Accepted 14 March 2017

KEYWORDSMyeloma; Infectiouscomplications; T-CellMediated Immunity; InnateImmunity

Introduction

Multiple myeloma (MM) is the second most frequenthematological malignancy. It derives from premalignantproliferation of monoclonal plasma cells and multistepgenetic and microenvironmental changes that results inmalignancy [1,2]. With the development of novel agentsand enhancement of supportive care, the five-year rela-tive survival rate has improved from 27.2% to 50.2% inthe past two decades [3]. MM and its monoclonal immu-noglobulins cause bone pain, renal failure, anemia, andincrease patients’ susceptibility to infections [4].

Mycobacteria tuberculosis (TB) remains an importantpublic health issue globally. It was responsible for 1.5million deaths in 2014 [5]. The risk of TB is increased inpatients with hematologic and head and neck neo-plasms [6]. The impaired immunity in multiple myeloma

may render myeloma patients more susceptible to TBinfections [7–13]. To the best of our knowledge, a fewstudies have examined the relationship between TB andMM but only at the level of subgroup analysis [14,15].

Taiwan is an endemic TB area and physicians hereare keenly aware of this disease, which offers a goodopportunity to study the relationship between MMand TB. The aims of the study were to examine therisk of TB infection in MM patients and the impact ofTB on MM patients.

Methods

Data sources

Taiwan’s National Health Insurance program waslaunched in 1995 to provide comprehensive healthcare

CONTACT Chia-Jen Liu chiajenliu@gmail.comJin-Hwang Liu jhliu@vghtpe.gov.tw Division of Hematology, Department of Medicine, TaipeiVeterans General Hospital, No. 201 Shipai Road, Sec. 2, Taipei 11217, Taiwan†Jin-Hwang Liu and Chia-Jen Liu contributed equally to the manuscript.�Institution at which the work was performed: Taipei Veterans General Hospital, Taipei, Taiwan

Supplemental data for this article can be accessed here.

� 2017 Informa UK Limited, trading as Taylor & Francis Group

LEUKEMIA & LYMPHOMA, 2017http://dx.doi.org/10.1080/10428194.2017.1312369

to all residents. Enrollment in this program is compul-sory and the coverage is universal (>99%). The pro-gram provides extensive medical care that includesinpatient, outpatient, emergency, dental care, as wellas traditional Chinese medicine services [16].

These multiple databases were released by Taiwan’sNational Health Research Institutes (NHRI) as theNational Health Insurance Research Database (NHIRD).The MM patients were identified from the Registry forCatastrophic Illness Patients (RCIP). All the registrationfiles and medical claims for this group from 1995onward, which has been primarily validated by theNHRI, and released in electronic form are included inthe database.

In this cohort database, original identifiable infor-mation from each patient has been encrypted in aconsistent cord, allowing linkage of individuals’ claims.Therefore, longitudinal follow-up of each patient’s his-tory of drug prescriptions and diagnosis of diseaseswithout reviewing their chart directly is feasible if thepatient has records from hospitals or clinics in Taiwanafter the year 2000. Because the information fromNHIRD is de-identified secondary data for research pur-poses, it is not necessary to obtain consent from thestudy patients, and the institutional review board ofTaipei Veterans General Hospital issued a formal writ-ten waiver for the need of consent (No. 2014-12-002AC).

Study population

MM patients were defined as patients with diagnosiscode 203 in the International Classification of Diseases,Ninth Revision, and Clinical Modification (ICD-9-CM)from our inpatient and outpatient imbursement sys-tem and conformed by RCIP to identify patients whohad been newly diagnosed with multiple myelomabetween 1 January 2000 and 31 December 2011. AllMM patients with pathological confirmation wereincluded in this study whether they received treatmentor not and whether their diseases were active orasymptomatic. Patients who were under age 20 atdiagnosis and who had a history of TB or confirmedTB diagnosis within 30 days were excluded. TB infec-tion is categorized as pulmonary and extrapulmonaryTB, according to WHO criteria [17]. Diagnosis of TBrequires laboratory and clinical diagnosis that isdefined by a compatible ICD-9-CM code (010–018).The diagnosis can be made by bacteriological confirm-ation or clinical diagnosis. According to WHO criteria, aclinically diagnosed TB case should be diagnosedwith active TB by a clinician or other medical practi-tioner who has decided to give the patient anti-TB

treatment [17]. We used a prescription of at least twoanti-TB medications for �28 days to confirm the diag-nosis. Pulmonary TB refers to any TB case involvingthe lung parenchyma or the tracheobronchial tree.Miliary TB is also classified as pulmonary TB.Extrapulmonary TB refers to any TB case involvingorgans other than lungs. A patient with both pulmon-ary and extrapulmonary TB was classified as a case ofpulmonary TB [17].

Matched cohort

We retrieved individuals without MM for the matchedcohort from the Longitudinal Health InsuranceDatabase 2000 (LHID2000), which consists of data on arandom sample of approximately one million peoplealive in 2000 to represent the Taiwanese population.Four non-multiple myeloma subjects were retrieved tomatch each myeloma patient by age, sex, and pres-ence of comorbidities on the same index date of rec-ognition of MM. We defined comorbidities as diseasesrelated to low immunity or lung disease that may pre-dispose patients to TB and be potential confoundingfactors [18–22]. The same exclusion criteria for thestudy cohort were applied to the matched cohort.

Statistical analysis

The main outcome of interest was the development ofTB. The two cohorts were followed until the occurrenceof TB, dropout from NHI, death, or the end of the studyperiod. Baseline characteristics were summarized as fre-quencies and percentages, medians and interquartileranges. Categorical variables were compared in categor-ical forms by using Chi-square tests. In the survival ana-lysis, the Kaplan–Meier method was used for theestimation of cumulative incidence of TB; differencesbetween the MM patients and matched cohort weretested by log-rank test. A multivariate Cox proportionalhazards model was used to identify risk factors for TBamong MM patients. Covariates such as age, sex, comor-bidities, and therapeutic modalities for multiple mye-loma were included in the model. Those with p< .1 inthe univariate model were used in the multivariate ana-lysis. Age-adjusted Charlson comorbidity index (ACCI)was incorporated in other multivariate analysis(Supplemental tables). All corticosteroid prescriptions inthe follow-up period were converted to the dailyequivalent dose of prednisolone. All treatments wereestimated with the use of time-dependent Cox propor-tional hazards models to prevent immortal time bias.

Extraction and computation of data were performedusing the Perl programing language (version 5.12.2).

2 C.-K. TSAI ET AL.

We used the Microsoft SQL Server 2012 (MicrosoftCorp., Redmond, WA) for data linkage, processing, andsampling. All statistical analysis was performed usingSAS 9.3 software (SAS Institute Inc., Cary, NC) or STATAstatistical software (version 12.1; StataCorp). A p< .05was considered statistically significant.

Results

Clinical characteristics of the study population

Figure 1 shows the patient selection flowchart. Fourthousand five hundred fifty-four patients with newlydiagnosed MM were identified. After excludingpatients who had undergone MM before age 20(n¼ 14), those with antecedent TB (n¼ 298), and thosediagnosed with TB or followed up for less than 30days (n¼ 256), 3979 patients were enrolled. Themedian age of the study population was 67 (interquar-tile range, 58–75 years) with a male-female ratio of1.37. The most common comorbidities were chronickidney disease (40.7%) and chronic obstructive pul-monary disease (COPD; 37.2%). Table 1 summarizesthe demographic characteristics and comorbidities ofpatients with MM and the matched cohort.

TB incidence rates

Eighty-three (2.1%) and 241 (1.5%) patients were diag-nosed with TB from the MM cohort and the matched

cohort, respectively. The incidence rate of TB was 95.5cases per 10,000 person-years of follow-up in the MMcohort and 32.9 cases per 10,000 person-years in thematched cohort. The Kaplan–Meier estimate revealedthat the cumulative incidence of TB was significantlyhigher among patients with MM than in the matchedcohort (p< .001) (Figure 2). After adjusting for age, sexand the comorbidities listed in Table 1, the MM cohortshowed a significant increased risk of developing TB(adjusted hazard ratio [HR] 3.11, 95% confidence inter-val [CI] 2.41–4.02, p< .001) (Table 2). Furthermore, MMpatients had increased risk for both pulmonary TB(adjusted HR 3.14, 95% CI 2.40–4.11, p< .001) andextrapulmonary TB (adjusted HR 2.74, 95% CI1.15–6.50, p¼ .023).

Risk factors for TB in patients with MM

Age, sex, and comorbidities were included for analysisin a Cox proportional hazards regression model. Inaddition to age and sex, the indicated risk factorsidentified in the univariate analysis, such as alcoholuse disorder, steroid, cytotoxic chemotherapy, andhematopoietic stem cell transplantation (HSCT), wereincluded in the multivariate analysis. Independent riskfactors were age �65 (adjusted HR 1.93, 95% CI1.19–3.15, p¼ .008), alcohol use disorder (adjusted HR2.86, 95% CI 1.24–6.62, p¼ .014), and a steroid daily

Figure 1. Patient selection flowchart.

TUBERCULOSIS INFECTION IN PATIENTS WITH MULTIPLE MYELOMA 3

dose equivalent to prednisone 5mg or more (adjustedHR 2.38, 95% CI 1.50–3.77, p< .001) (Table 3). Wereplaced the comorbidities with ACCI in multivariateanalysis as an alternative way of adjustment.Supplemental Table 1(A) shows that both MM andACCI are risk factors for TB infection in the multivariateanalysis. The adjusted HR were 2.63 (95% CI 2.04–3.39,p< .001) and 1.12 (95% CI 1.08–1.16, p< .001)(Supplemental Table 1(A)). The result was consistentwith our previous analysis adjusting many comorbid-ities. Moreover, MM was still an independent risk fac-tor for TB infection when the two cohorts were

stratified according to ACCI score except for the groupwith ACCI score �9, which might render MM a lesssignificant risk factor (Supplemental Table 1(B)).

Relationship between steroid dose and risk of TBin patients with MM

Most of the MM patients (98.6%) received steroidtreatment after diagnosis. We calculated the steroiddose as a prednisolone-equivalent daily dose. In theunivariate analysis, the crude HR for daily use of0–5mg, 5–10mg, and �10mg of prednisolone or itsequivalents were 1.71 (95% CI 0.95–3.08, p¼ .075),2.19 (95% CI 1.08–4.48, p¼ .031), and 3.07 (95% CI1.76–5.35, p< .001). In the multivariate analysis, theadjusted HR for daily use of 0–5mg, 5–10mg, and�10mg of prednisolone or its equivalents comparedwith no use were 1.46 (95% CI 0.80–2.66, p¼ .221),2.00 (95% CI 0.97–4.11, p¼ .060), and 3.23 (95% CI1.84–5.69, p< .001) (Table 4). There was also a positivedose–response relationship between steroid dose andrisk of TB (adjusted p for trend< .001).

Analysis of outcome of MM patients with TB

Among the MM cohort, 147 (45.4%) patients died dur-ing the study period. By treating TB development as atime-dependent covariate in the Cox proportional haz-ards model, MM patients with TB had a higher

Table 1. Baseline characteristics of patients with MM and the matched cohort.Patients with MM

n¼ 3979Matched cohort

n¼ 15,916

Characteristics n % n % p Value

Median age, years (interquartile range) 67 (58–75) 67 (58–75)�65 2301 57.8 9,204 57.8 1.000<65 1678 42.2 6,712 42.2

SexMale 2300 57.8 9,200 57.8 1.000Female 1679 42.2 6,716 42.2 .

ComorbiditiesDiabetes mellitus 1241 31.2 4966 31.2 .988COPD 1479 37.2 5917 37.2 .994Asthma 850 21.4 3398 21.3 .986Chronic kidney disease 1620 40.7 6477 40.7 .983Alcohol use disorder 109 2.7 434 2.7 .965Liver cirrhosis 200 5.0 799 5.0 .987Autoimmune diseases 521 13.1 2082 13.1 .983Dyslipidemia 1389 34.9 5555 34.9 .994Gastrectomy 26 0.7 106 0.7 .930Silicosis 3 0.1 24 0.2 .248HIV 9 0.2 32 0.2 .755

TreatmentBortezomib 721 18.1Thalidomide 1379 34.7Steroid 3924 98.6Cytotoxic chemotherapy 3277 82.4HSCT 342 8.6

MM: multiple myeloma; COPD: chronic obstructive pulmonary disease; HSCT: hematopoietic stem cell transplantation; HIV: humanimmunodeficiency virus.

Figure 2. Cumulative incidence of TB in patients with MM andthe matched cohort.

4 C.-K. TSAI ET AL.

mortality rate than those without TB (adjusted HR2.03, 95% CI 1.54–2.67, p< .001) after adjusting forother variables, such as age �65, sex, diabetes melli-tus, COPD, asthma, chronic kidney disease, and liver

cirrhosis (Table 5). We also incorporated ACCI in theunivariate analysis and multivariate analysis of risk fac-tors for mortality in Supplemental Table 2. TB is still acritical risk factor for mortality (adjusted HR 2.29, 95%CI 1.74–3.01, p< .001) as well as ACCI score (adjustedHR 1.08, 95% CI 1.07–1.10, p< .001).

Discussion

This nationwide population-based study shows thatMM patients have a higher risk of subsequent TBinfection, including both pulmonary and extrapulmo-nary TB. Furthermore, our MM patients with TB infec-tion had a twice-higher overall mortality rate thanthose without. Our multivariate analysis shows thatindependent risk factorsof TB infection for MM

Table 2. Incidence of TB in patients with MM and the matched cohort.Patients with MM Matched cohort

TB no.Per 10,000person-years TB no.

Per 10,000person-years Crude HR (95% CI) p Value Adjusted HRa (95% CI) p Value

Total 83 95.5 241 32.9 2.78 (2.16–3.59) <.001 3.11 (2.41–4.02) <.001Age

�65 53 130.6 186 47.5 2.64 (1.93–3.61) <.001 2.75 (2.01–3.76) <.001<65 30 64.7 55 16.1 4.03 (2.56–6.34) <.001 4.14 (2.62–6.53) <.001

SexMale 57 114.4 196 46.2 2.40 (1.78–3.25) <.001 2.70 (2.00–3.66) <.001Female 26 70.0 45 14.6 4.41 (2.69–7.23) <.001 4.81 (2.92–7.93) <.001

TypePulmonary TB 76 87.4 218 29.8 2.82 (2.16–3.68) <.001 3.14 (2.40–4.11) <.001Extrapulmonary TB 7 8.1 23 3.1 2.44 (1.03–5.75) .042 2.74 (1.15–6.50) .023

TB: tuberculosis; MM: multiple myeloma; HR: hazard ratio; CI: confidence interval.aAdjusted for age, sex, and the comorbidities listed in Table 1.

Table 3. Analysis of risk factors for TB in patients with MM.Univariate analysis Multivariate analysisa

Predictive variables HR (95% CI) p Value HR (95% CI) p Value

Age �65 1.94 (1.23–3.05) .004 1.93 (1.19–3.15) .008Male sex 1.64 (1.03–2.60) .038 1.55 (0.97–2.47) .066Comorbidities

Diabetes mellitus 0.81 (0.49–1.36) .428COPD 1.18 (0.75–1.84) .473Asthma 1.45 (0.88–2.37) .146Chronic kidney disease 1.03 (0.65–1.63) .903Alcohol use disorder 3.00 (1.31–6.89) .010 2.86 (1.24–6.62) .014Liver cirrhosis 0.50 (0.12–2.04) .336Autoimmune diseases 0.87 (0.43–1.73) .683Dyslipidemia 1.07 (0.68–1.68) .780Gastrectomy d

Silicosis d

HIV d

Treatmentb

Bortezomib 1.12 (0.51–2.47) .780Thalidomide 0.82 (0.45–1.48) .511Steroidc 2.31 (1.46–3.65) <.001 2.38 (1.50–3.77) <.001Cytotoxic chemotherapy 1.76 (0.93–3.33) .083 1.55 (0.81–2.96) .185HSCT 0.48 (0.21–1.11) .087 0.62 (0.25–1.51) .289

TB: tuberculosis; MM: multiple myeloma; HR: hazard ratio; CI: confidence interval; COPD: chronic obstructive pulmonary disease;HSCT: hematopoietic stem cell transplantation; HIV: human immunodeficiency virus.aAll factors with p< .1 in the univariate analysis were included in the Cox multivariate analysis.bTreatment was analyzed as a time-dependent covariate in the Cox regression model.cRepresents steroid dose equivalent to prednisone 5mg or more daily and was analyzed as a time-dependent covariate in the Coxregression model.dDo not converge.

Table 4. Relationship between steroid dose and risk of TB inpatients with MM.

Univariate analysis Multivariate analysisa

Exposure HR (95% CI) p Value HR (95% CI) p Value

Steroid daily doseb

No use reference reference0–5mg 1.71 (0.95–3.08) .075 1.46 (0.80–2.66) .2215–10mg 2.19 (1.08–4.48) .031 2.00 (0.97–4.11) .060� 10mg 3.07 (1.76–5.35) <.001 3.23 (1.84–5.69) <.001

TB: tuberculosis; MM: multiple myeloma; HR: hazard ratio; CI: confidenceinterval.aAdjusted for age, sex, alcohol use disorder, and Cytotoxic chemotherapy,and hematopoietic stem cell transplantation.bRepresents the daily dose equivalent of prednisone and was analyzed asa time-dependent covariate in the Cox regression model.

TUBERCULOSIS INFECTION IN PATIENTS WITH MULTIPLE MYELOMA 5

patients included older age (� 65 years), alcohol usedisorder, and steroid treatment (dose equivalent toprednisone �5mg daily).

A previous single-center retrospective study inTaiwan reported TB prevalence in patients with hema-tological malignancies and MM were 1.78% and 1.63%,respectively [14], which is lower than the prevalence inour study. Our population-based study, with longerfollow-up period and a larger case number, might becloser to the true incidence in comparison to the pre-vious study.

TB is a risk factor for mortality in patients with mul-tiple myeloma (adjusted HR 2.03, p< .001). Studieshave revealed that developing TB results in discontinu-ation of chemotherapy, delay in surgery, and even mor-tality [7]. Moreover, anti-TB treatment might be abateddue to coexisting malignancies [23–25]. Our studyshows that MM patients with TB had lower overall sur-vival than those without. In addition to TB, being male,having DM and CKD were also risk factors for mortality.A study including 1960 patients from a trial reportedfemale gender was associated with inferior overall sur-vival than male gender [26]. The increased prevalenceof adverse genetic lesions in female myeloma patientsmight be the cause. However, a population-based studyabout newly diagnosed MM patients pointed outfemale sex was associated with decreased risk of death(HR 0.86, 95% CI 0.75–0.98) [27]. Another retrospectivecohort study reported the inpatient mortality of menwas 1.162 times that of women (p¼ .012) [28]. Anotherpopulation based study reported male sex was one ofthe risk factors for early mortality in patients with MM(female to male odds ratio (OR) 0.90, 95% CI 0.85–0.95)[29]. A previous single-center retrospective study inTaiwan reported being male (adjusted OR 2.93, 95% CI1.17–7.31) was a risk factor of early mortality (deathwithin 60 days after diagnosis) in patients with newly

diagnosed MM [30]. In the aforementioned Taiwanstudy, male patients were older (p¼ .01), had higherserum creatinine (p¼ .028), lower serum albumin(p¼ .022), and more advanced ECOG stages (p¼ .043)compared with female patients. Being male, afteradjusting for the above factors, remained significant forearly mortality. This suggests the mortality differencemight be multifactorial. Our result was compatible withthese previous studies. There is also a descriptive studycontaining 217 Egyptian myeloma patients showing nodifference in survival between different genders [31].Further research is warranted to elucidate the problem.

Advanced age [18], being male, and having DM[19], COPD [20], liver cirrhosis [21], and end-stage renaldisease [22] are known risk factors for TB in the gen-eral population. The study adjusted for the potentialconfounders according to literature reviews and showsonly older age, alcohol use disorder, and steroid treat-ment to be independent risk factors in our MM cohort.TB incidence also increased progressively with age in aWHO global report [5]. The present study also showsthat advanced age is an independent risk factor in MMpatients.

A nationwide study using data reported to theNational Tuberculosis Surveillance System in the US in1997–2012 reported excessive alcohol use was inde-pendently associated with TB infection, diagnosed viaboth positive AFB sputum smear microscopy results(adjusted odds ratio [OR] 1.23, 95% CI 1.18–1.28) andpositive culture results (adjusted OR 1.36, 95% CI1.29–1.43) [32]. A meta-analysis based on three cohortsand 18 case–control studies disclosed a pooled relativerisk of 2.94 for alcohol use as a risk factor for active TB[33]. Alcohol use disorder has been associated with animpaired immune system and predisposes patients tovaried infection, especially pulmonary infection [34,35].A TB treatment trial conducted in North America

Table 5. Risk factors for mortality in patients with MM.Univariate analysis Multivariate analysisa

Predictive variables HR (95% CI) p Value HR (95% CI) p Value

Tuberculosisb 2.32 (1.77–3.05) <.001 2.03 (1.54–2.67) <.001Age �65 1.81 (1.65–1.98) <.001 1.12 (1.03–1.23) .010Male sex 1.14 (1.04–1.24) .004 1.71 (1.56–1.88) <.001ComorbiditiesDiabetes mellitus 1.26 (1.15–1.38) <.001 1.11 (1.01–1.22) .034COPD 1.25 (1.15–1.37) <.001 1.06 (0.96–1.16) .245Asthma 1.12 (1.01–1.25) .035 0.98 (0.88–1.10) .761Chronic kidney disease 1.46 (1.34–1.59) <.001 1.38 (1.26–1.51) <.001Alcohol use disorder 1.04 (0.80–1.36) .759Liver cirrhosis 1.22 (1.01–1.47) .039 1.15 (0.95–1.38) .154Autoimmune diseases 1.03 (0.90–1.17) .665Dyslipidemia 0.98 (0.90–1.00) .683

MM: multiple myeloma; HR: hazard ratio; CI: confidence interval; COPD: chronic obstructive pulmonary disease.aAll factors with p< 0.1 in the univariate analysis were included in the Cox multivariate analysis.bTuberculosis was analyzed as a time-dependent covariate in the Cox regression model.

6 C.-K. TSAI ET AL.

containing 1075 patients identified alcohol use as anindependent risk factor for death [36]. Our resultsdemonstrated a higher rate of TB infection in patientswith alcohol use disorder but alcohol use disorderitself was not associated with higher mortality.

Glucocorticoid use is believed to increase the risk ofTB infection. A case–control study of 497 TB casesrevealed increased risk of TB infection compared withcontrols without use (adjusted OR 4.9), with adose–response relationship [37]. Our study demon-strates that the risk of TB infection in MM patientsincreased significantly in those with a prednisoloneequivalent daily dose of �5mg. Similar to previousstudies, it also demonstrates a dose-dependent effect.

Chemotherapy increases the risk of developing TB[38]. Fan [15] and Arslen [39] et al. reported that HSCTis also associated with an increased risk of TB inendemic regions. However, HSCT, bortezomib, thalido-mide, and cytotoxic chemotherapy, were not inde-pendent risk factors for TB development in our study.However, MM patients receiving cytotoxic chemother-apy had a trend of increased risk of TB infection.

The increased incidence of extrapulmonary TB inmyeloma patients (adjusted HR 2.74) may be the mani-festation of impaired immunity against the dissemin-ation of TB [40]. The immunity impairment in multiplemyeloma includes hypogammaglobulinemia, impaireddendritic cell number and function [9,11], change of Tcell composition such as increased number of CD4 Tregulatory (Treg) cells, CD8 Treg cells resulting in sup-pressing the effector T cells [12], etc. The defectiveinnate and adaptive immune responses lead to thesusceptibility to TB infection [13].

Our findings are valuable for the objective analysis ofTB in patients with multiple myeloma owing to the largesample size, unbiased subject selection, and strict diag-nostic criteria. Certification of multiple myeloma, as oneof the NHI-defined catastrophic illness, can exemptpatients from related medical expenses and requiresstrict verification process. Additionally, TB is a notifiabledisease in Taiwan and, therefore, by law must bereported to the CDC within 1 week of diagnosis. In add-ition to prescription of anti-TB medications for �28 daysto confirm diagnosis, the diagnosis of both multiplemyeloma and TB in our study is highly reliable.

This study has limitations. First, the administrativedata used in this study did not have information onsmoking status, socioeconomic status, performancestatus, laboratory data like lactate dehydrogenase(LDH) and albumin, staging of multiple myeloma suchas International Staging System (ISS), or cytogenetics,which should be included in univariate and multivari-ate analysis. As a result, this study cannot examine the

relationship between TB and these parameters indetail. Second, we could not differentiate whether thediagnosis of TB was based on bacteriological or clinicaldiagnosis in the insurance registries. In Taiwan, how-ever, diagnosis of TB should be reported to govern-ment. Therefore, the diagnosis is highly reliable. Third,we excluded patients who had antecedent TB beforeMM diagnosis to clarify the relationship between MMand newly diagnosed TB because it was hard to differ-entiate whether TB infection was a result of reactiva-tion of old foci or new infections.

Conclusion

The risk of TB, both pulmonary and extrapulmonary, issignificantly higher in myeloma patients. Myelomapatients who contracted TB had a higher mortalityrate than those who did not. Therefore, judiciousscreening strategies for TB should be considered formyeloma patients with risk factors, such as old age,alcohol use disorder, and steroid treatment.

Acknowledgements

This study is based on data from the National HealthInsurance Research Database, provided by the Bureau ofNational Health Insurance, Department of Health, and man-aged by the National Health Research Institutes. The interpret-ation and conclusions contained herein do not representthose of the Bureau of National Health Insurance, Departmentof Health, or the National Health Research Institutes.

Potential conflict of interest: Disclosure forms providedby the authors are available with the full text of this articleonline at http://dx.doi.org/10.1080/10428194.2017.1312369.

Funding

This study was supported by grants from Taipei VeteransGeneral Hospital [V105B-016 and V105E10-002-MY2-1], theMinistry of Science and Technology [MOST 104-2314-B-075-085-MY2], the Taiwan Clinical Oncology Research Foundation,Szu-Yuan Research Foundation of Internal Medicine, andChong Hin Loon Memorial Cancer and Biotherapy ResearchCenter at National Yang-Ming University.

ORCID

Chun-Kuang Tsai http://orcid.org/0000-0002-8227-7585

References

[1] Rollig C, Knop S, Bornhauser M. Multiple myeloma.Lancet. 2015;385:2197–2208.

TUBERCULOSIS INFECTION IN PATIENTS WITH MULTIPLE MYELOMA 7

[2] Palumbo A, Anderson K. Multiple myeloma. N Engl JMed. 2011;364:1046–1060.

[3] Howlader N, Noone AM, Krapcho M, et al (eds). SEERCancer Statistics Review, 1975–2013 [cited 2016 April].Bethesda, MD: National Cancer Institute. Availablefrom: http://seer.cancer.gov/csr/1975_2013/, based onNovember 2015 SEER data submission.

[4] Munshi NC, Longo DL, Anderson KC, Plasma cell disor-ders. In: Kasper DL, Fauci AS, Longo DL, Hauser SL,Jameson JL, Loscalzo J. Harrison's principles ofinternal medicine. 19th ed. Vol 1. New York: McGraw-Hill; 2015. p. 712.

[5] Laura A, Anna D, Dennis F, et al. Global tuberculosisreport. 20th ed. Geneva, Switzerland: World HealthOrganization; 2015, p. 1.

[6] Kamboj M, Sepkowitz KA. The risk of tuberculosis inpatients with cancer. Clin Infect Dis. 2006;42:1592–1595.

[7] Ahn JS, Rew SY, Yang DH, et al. Poor prognostic sig-nificance of Mycobacterium tuberculosis infection dur-ing bortezomib-containing chemotherapy in patientswith multiple myeloma. Blood Res. 2013;48:35–39.

[8] Mills KH, Cawley JC. Abnormal monoclonal antibody-defined helper/suppressor T-cell subpopulations inmultiple myeloma: relationship to treatment and clin-ical stage. Br J Haematol. 1983;53:271–275.

[9] Brown RD, Pope B, Murray A, et al. Dendritic cellsfrom patients with myeloma are numerically normalbut functionally defective as they fail to up-regulateCD80 (B7-1) expression after huCD40LT stimulationbecause of inhibition by transforming growth factor-beta1 and interleukin-10. Blood. 2001;98:2992–2998.

[10] Ogawara H, Handa H, Yamazaki T, et al. High Th1/Th2ratio in patients with multiple myeloma. Leuk Res.2005;29:135–140.

[11] Ratta M, Fagnoni F, Curti A, et al. Dendritic cells arefunctionally defective in multiple myeloma: the role ofinterleukin-6. Blood. 2002;100:230–237.

[12] Muthu Raja KR, Kubiczkova L, Rihova L, et al.Functionally suppressive CD8 T regulatory cells areincreased in patients with multiple myeloma: a causefor immune impairment. PLoS One. 2012;7:e49446.

[13] Philips JA, Ernst JD. Tuberculosis pathogenesis andimmunity. Annu Rev Pathol. 2012;7:353–384.

[14] Chien-Yuan Chen W-HS, Cheng A, Tsay W, Huang S-Y,Tang J-L, Chen Y-C, Wang J-Y, Tien H-F, Chang S-C.Clinical characteristics and outcomes ofMycobacterium tuberculosis disease in adult patients.BMC Infect Dis. 2011;11:324.

[15] Fan WC, Liu CJ, Hong YC, et al. Long-term risk oftuberculosis in haematopoietic stem cell transplantrecipients: a 10-year nationwide study. Int J TubercLung Dis. 2015;19:58–64.

[16] Cheng TM. Taiwan's new national health insuranceprogram: genesis and experience so far. Health Aff(Millwood). 2003;22:61–76.

[17] World Health Organization. (2013, Apr 15). Definitionsand reporting framework for tuberculosis. Availablefrom: http://apps.who.int/iris/bitstream/10665/79199/1/9789241505345_eng.pdf.

[18] Stead WW, Lofgren JP. Does the risk of tuberculosisincrease in old age? J Infect Dis. 1983;147:951–955.

[19] Dooley KE, Chaisson RE. Tuberculosis and diabetesmellitus: convergence of two epidemics. Lancet InfectDis. 2009;9:737–746.

[20] Lee CH, Lee MC, Shu CC, et al. Risk factors for pul-monary tuberculosis in patients with chronic obstruct-ive airway disease in Taiwan: a nationwide cohortstudy. BMC Infect Dis. 2013;13:194.

[21] Cho YJ, Lee SM, Yoo CG, et al. Clinical characteristicsof tuberculosis in patients with liver cirrhosis.Respirology. 2007;12:401–405.

[22] Hu HY, Wu CY, Huang N, et al. Increased risk of tuber-culosis in patients with end-stage renal disease: apopulation-based cohort study in Taiwan, a countryof high incidence of end-stage renal disease.Epidemiol Infect. 2014;142:191–199.

[23] Fujita T, Endo M, Gu Y, et al. Mycobacterium tubercu-losis infection in cancer patients at a tertiary care can-cer center in Japan. J Infect Chemother. 2014;20:213–216.

[24] Kim DK, Lee SW, Yoo CG, et al. Clinical characteristicsand treatment responses of tuberculosis in patientswith malignancy receiving anticancer chemotherapy.Chest. 2005;128:2218–2222.

[25] De La Rosa GR, Jacobson KL, Rolston KV, et al.Mycobacterium tuberculosis at a comprehensive can-cer centre: active disease in patients with underlyingmalignancy during 1990–2000. Clin Microbiol Infect.2004;10:749–752.

[26] Boyd KD, Ross FM, Chiecchio L, et al. Gender dispar-ities in the tumor genetics and clinical outcome ofmultiple myeloma. Cancer Epidemiol Biomarkers Prev.2011;20:1703–1707.

[27] Yusuf AA, Natwick T, Werther W, et al. A retrospect-ive analysis to examine factors associated with mor-tality in Medicare beneficiaries newly diagnosedwith multiple myeloma. Curr Med Res Opin.2016;32:1989–1996.

[28] Chen JH, Chung CH, Wang YC, et al. Prevalence andmortality-related factors of multiple myeloma inTaiwan. PLoS One. 2016;11:e0167227.

[29] Costa LJ, Gonsalves WI, Kumar SK. Earlymortality in multiple myeloma. Leukemia. 2015;29:1616–1618.

[30] Hsu P, Lin TW, Gau JP, et al. Risk of EarlyMortality in Patients With Newly DiagnosedMultiple Myeloma. Medicine (Baltimore). 2015;94:e2305.

[31] El Husseiny NM, Kasem N, El Azeeim HA, et al.Multiple myeloma: a descriptive study of 217 Egyptianpatients. Ann Hematol. 2014;93:141–145.

[32] Volkmann T, Moonan PK, Miramontes R, et al.Tuberculosis and excess alcohol use in the UnitedStates, 1997-2012. Int J Tuberc Lung Dis. 2015;19:111–119.

[33] Rehm J, Samokhvalov AV, Neuman MG, et al. Theassociation between alcohol use, alcohol use disordersand tuberculosis (TB). A systematic review. BMC PublicHealth. 2009;9:450.

[34] Zhang P, Bagby GJ, Happel KI, et al. Pulmonary hostdefenses and alcohol. Front Biosci. 2002;7:d1314–d1330.

8 C.-K. TSAI ET AL.

[35] Happel KI, Nelson S. Alcohol, immunosuppression,and the lung. Proc Am Thorac Soc. 2005;2:428–432.

[36] Sterling TR, Zhao Z, Khan A, et al. Mortality in a largetuberculosis treatment trial: modifiable and non-modi-fiable risk factors. Int J Tuberc Lung Dis. 2006;10:542–549.

[37] Jick SS, Lieberman ES, Rahman MU, et al.Glucocorticoid use, other associated factors, and therisk of tuberculosis. Arthritis Rheum. 2006;55:19–26.

[38] Kim HR, Hwang SS, Ro YK, et al. Solid-organ malig-nancy as a risk factor for tuberculosis. Respirology.2008;13:413–419.

[39] Arslan O, Gurman G, Dilek I, et al. Incidence of tubercu-losis after bone marrow transplantation in a single cen-ter from Turkey. Haematologia (Budap). 1998;29:59–62.

[40] Krishnan N, Robertson BD, Thwaites G. The mecha-nisms and consequences of the extra-pulmonarydissemination of Mycobacterium tuberculosis.Tuberculosis (Edinb). 2010;90:361–366.

TUBERCULOSIS INFECTION IN PATIENTS WITH MULTIPLE MYELOMA 9