Surgical Resection Compared to Medical Management in localized Mycobacteria Avium Complex Pulmonary Infection s-A Case Control Study

Pulmonary infections associated with Mycobacterium avium complex can be challenging to treat medically and the role of surgical lung resection is not well established. We aim to assess safety and microbiologic response in patient with localized Mycobacterium avium complex pulmonary infections managed with surgical lung resection compared to medical management alone. We present a multi-institutional case series of 16 patients with localized Mycobacterium avium complex pulmonary infections managed with surgical lung resection. We highlight the case of a 30 to 40-year-old patient with localized pulmonary disease amenable to surgical resection for illustrative purposes and report on outcomes compared with medically treated patients at the same institution in case-control design. Of 745 patients meeting microbiologic diagnostic criteria for Mycobacterium avium complex pulmonary infections, 98 had localized pulmonary disease and of these 16 underwent surgical resection. Univariate and multivariate analysis results indicated no difference in surgical resection group compared with medical treatment: microbiologic response rate (odds ratio 0.49, 0.1–2.41), 2-year all-cause mortality (odds ratio 0.87, 0.18–4.32), and composite outcome of 2-year mortality and lack of microbiological response (multivariate logistic regression OR = 0.45, 0.09–1.57). rarely 0.18-4.32) between surgical and medical cohorts. The composite outcome of 2-year mortality and lack of microbiological response showed no signicant difference between the two groups after adjustment for age and gender (multivariate logistic regression OR=0.45, 0.09-1.57). localized with clinical and microbiologic signicant, the measured outcome odds ratios were trending towards favoring surgery across microbiological response and survival. Our institutional experience is limited by low numbers of patients with MAC that underwent surgical lung resection. Surgical lung resection in patients with NTM at high volume specialty centers is associated with low rates of associated morbidity and mortality 9 . Our case series adds data from an intermediate volume health system and also reports on microbiological clearance and overall outcome compared with medically treated patients. We highlight the case of a 30 to 40-year-old patient with MAC pulmonary infection localized to the left upper lobe. Her course demonstrates the diculty in treating MAC pulmonary infection despite long courses of multiple anti-mycobacterial agents. Though, she had negative sputum cultures after initiation of medical therapy, the persistence of symptoms and cavitation raised concerns about a sustained microbiological clearance, with the cavitary lesion acting a reservoir for disease. In particular, her case highlights that patients with well localized disease with few comorbidities may benet from surgical resection compared to medical therapy alone.


Abstract
Background Pulmonary infections associated with Mycobacterium avium complex can be challenging to treat medically and the role of surgical lung resection is not well established. We aim to assess safety and microbiologic response in patient with localized Mycobacterium avium complex pulmonary infections managed with surgical lung resection compared to medical management alone.

Methods
We present a multi-institutional case series of 16 patients with localized Mycobacterium avium complex pulmonary infections managed with surgical lung resection. We highlight the case of a 30 to 40-year-old patient with localized pulmonary disease amenable to surgical resection for illustrative purposes and report on outcomes compared with medically treated patients at the same institution in case-control design.

Results
Of 745 patients meeting microbiologic diagnostic criteria for Mycobacterium avium complex pulmonary infections, 98 had localized pulmonary disease and of these 16 underwent surgical resection. Univariate and multivariate analysis results indicated no difference in surgical resection group compared with medical treatment: microbiologic response rate (odds ratio 0.49, 0.1-2.41), 2-year all-cause mortality (odds ratio 0.87, 0.18-4.32), and composite outcome of 2-year mortality and lack of microbiological response (multivariate logistic regression OR = 0.45, 0.09-1.57).

Conclusions
This case series describes patients with localized pulmonary Mycobacterium avium complex for whom surgical resection was pursued and shows examples of patients that may bene t from surgery. Though surgery for pulmonary Mycobacterium avium complex disease is rarely performed, it appears as safe and at least as effective as medical-therapy alone.

Background
Mycobacterium avium complex (MAC) pulmonary infections are increasing in prevalence in the US and are associated with high rates of morbidity, mortality and considerable health care cost 1-3 . Medical treatment is complex and consists of 3-5 antibiotics administered for 12-18 months or longer, yet clinical and microbiological response rates are limited and range as low as 55% 4,5 . As a result, for patients with anatomically appropriate disease, surgical resection has been proposed as adjuvant therapy, but there is limited data on safety and treatment outcomes in intermediate volume centers 6 . There has been one case-control study and several case series describing outcomes of lung resection for Non-tuberculous mycobacterial pulmonary infection (NTMI) that have mostly focused on the more virulent Mycobacterium abscessus 7,8 .

Methods
We conducted a retrospective review using an electronic health record database of all patients treated between 2001-2016 across Partners Healthcare, a multi-institutional health care system located in Boston and its suburbs. We identi ed all patients with ICD-9 (31.0/31.2) or ICD10 (A31.0/A31.2) diagnostic codes for NTMI, and identi ed patients with the presence of ≥ 2 expectorated respiratory samples or a ≥ 1 bronchioalveolar lavage sample positive for NTM species meeting ATS microbiologic criteria for diagnosis of NTMI 1 . Using natural language processing with regular expressions, we extracted from our electronic health record basic demographics, anthropomorphic data, comorbidities, pulmonary function measurement, radiologic reports, microbiologic data, and surgical reports to build a database of patients. We restricted to patients infected with MAC, had radiographically localized mycobacterial infection, and received medical treatment for MAC during their treatment course. Anatomically limited/localized disease appropriate for surgical resection was de ned as radiologic evidence for lesions typical of NTMI (bronchiectasis, nodular opacities, or consolidation) isolated to one lobe or any patient with a cavitary lesion regardless of number of lobes affected. Adequate MAC treatment was de ned as a drug regimen containing at least the following drugs: (1) rst line: macrolide, ethambutol, and rifampicin or other rifamycin, or (2) second line: rst line with substitution of any of the drugs for clofazimine, bedaquline, or amikacin as these are often used as escalation therapy or in cases of drug intolerance. We assessed the safety and effect of surgical resection on treatment outcomes in patients with MAC pulmonary infections univariate outcomes and a logistic regression model using a composite variable in a case control design comparing patients who underwent surgical resection compared to medical therapy alone.
Outcome measures were de ned as microbiologic response and mortality. Microbiological conversion was de ned as either of the two following criteria met within two years of follow up (1) ≥ 2 negative cultures for patients with two or more follow-up cultures available without any recurrent positive culture or (2) one negative culture for patients with only one follow-up culture available. Mortality was de ned as 2-year all-cause mortality from the date of rst diagnosis. We additionally de ned a composite outcome of 2-year mortality and lack of microbiological response, and adjusted for age and gender using multivariate logistic regression. Approval for this study was obtained from the Partners Health System Institutional Review Board (2017P002469/PHS).

Results
In our database, there were 745 patients meeting ATS criteria for diagnosis of atypical mycobacterial pulmonary infection, of which 554 (74%) were infected with MAC based on available microbiologic data. A total of 98 patients met criteria for radiographically localized MAC infection and received medical treatment. Of these, 16 underwent surgical resection. Table 1 outlines demographics, comorbidities, medical treatment administered, and information regarding surgical resection. Table 2 outlines complications and outcomes for 16 surgically treated patients. Overall, surgical resection was performed rarely: in 16/554 (2.9%) of MAC patients meeting ATS criteria and 16/98 (16%) of the subset with localized disease requiring medical treatment. Of the 16 surgically treated patients, 13 underwent one lung resection. The remaining three patients underwent two resections at separate time intervals for progression of disease. The majority 10/16 (63%) of patients had surgical resection of disease from one lobe and the other six patients had a multi-lobar resection. The indication for surgery was listed as disease refractory to medical therapy in 9/16 (56%) patients and in 6/16 (38%) surgical resection was described as adjunct to medical therapy. For the remaining patient, the indication for surgery was described as hemoptysis with perioperative antimycobacterial administration. Eleven patients had cavitary lesions resected, 4 had bronchiectasis resected, 2 had consolidation resected, and 1 had an abscess resected. Postoperatively there were no bleeding events requiring repeat surgery. Only one of the 16 patients developed a bronchopleural stula (6%) and another developed a chyle leak (6%).
The baseline characteristics of the surgically treated groups are compared to the medically treated group with disease in Table 2. In univariate analysis on treatment outcome there was no signi cant difference in microbiologic response rate (odds ratio 0.49, 0.1-2.41) and 2-year all-cause mortality (odds ratio 0.87, 0.18-4.32) between surgical and medical cohorts. The composite outcome of 2-year mortality and lack of microbiological response showed no signi cant difference between the two groups after adjustment for age and gender (multivariate logistic regression OR=0.45, 0.09-1.57).

Brief Case Presentation
A 30 to 40-year-old patient with a 25 pack-year history of smoking, hypertension, seasonal allergies, and gastroesophageal re ux disease presented with 5-6 months of night sweats, associated weight loss, and chronic productive cough with brown sputum. Prior evaluations for these symptoms about a month prior to presentation were notable for chest radiograph initially visualizing left upper lobe opacities, but interval imaging at the time of presentation showed progression to left upper lobe cavitary lesion. Subsequently, this was con rmed on computed tomography of the chest with revisualization of left upper lobe cavitary lesion, with multiple satellite nodules. Initial induced sputum was notable for abundant 3-4+ acid fast bacilli on two subsequent days. Follow-up bronchoscopy and bronchioalveolar lavage with cultures con rmed a diagnosis of Mycobacterium Avium Complex pulmonary infection. Additional etiologies of cavitary lesions were excluded with negative tuberculin skin testing, negative coccidiosis serum titers, and negative ANCA testing. Susceptibility testing showed MAC susceptibility to clarithromycin.
About a month after initial presentation the decision was made to initiate medical therapy with Azithromycin, Ethambutol, and Rifampin. She tolerated medical therapy well with no hearing problems, no tinnitus, no imbalance, no visual changes, no color vision problems, no nausea, no vomiting, and no diarrhea. About 2 months after initiation of medical therapy surveillance mycobacterial cultures were negative. Initial surveillance chest computed tomography 4 months after therapy initiation showed an interval response to therapy, marked by decrease in the number of satellite nodules and improvement in the cavitary wall thickness and reduced in ammation. Interval imaging 11 months after initiation of therapy, however, showed a persistence of the cavitary lesion. Given persistent unresolved symptoms and the radiographic ndings she was referred for surgical evaluation.
With well localized disease, she was deemed a candidate for limited parenchymal surgical resection with curative intent. She underwent a surgical wedge resection of the left upper lobe cavitary 14 months after therapy initiation without any post-operative complications. She received 4 weeks of preoperative and postoperative intravenous amikacin. On pathologic examination, the cavitary lesion was tan-white with a rm wall containing necrotic debris; margins were 0.4 cm and the uninvolved parenchyma was normal without masses. Mycobacterial organisms were visible with Acid fast stain.
In addition to 4 weeks of post-operative intravenous amikacin, a plan was made to continue Azithromycin, Ethambutol, and Rifampin for 1 year postoperatively, but the patient terminated therapy at 6 months for non-medical reasons. Intermittent surveillance cultures were obtained for 16 months after resection remained negative. At the time of follow-up 16 months of after resection she remained asymptomatic without cough, night sweats, or weight loss.

Discussion
We systematically identify a series of surgically treated patients with localized MAC pulmonary disease in an intermediate volume multi-institutional tertiary health care system and present the rst case-control study assessing safety and outcomes after surgical lung resection for MAC disease, with prior studies focusing on Mycobacterium abscessus 7,8 . We nd the rate of surgical complications to be low (≤ 6%) suggesting that resection of localized MAC disease is safe and may be associated with improved clinical and microbiologic response. Though not statistically signi cant, the measured outcome odds ratios were trending towards favoring surgery across microbiological response and survival.
Our institutional experience is limited by low numbers of patients with MAC that underwent surgical lung resection. Surgical lung resection in patients with NTM at high volume specialty centers is associated with low rates of associated morbidity and mortality 9 . Our case series adds data from an intermediate volume health system and also reports on microbiological clearance and overall outcome compared with medically treated patients.
We highlight the case of a 30 to 40-year-old patient with MAC pulmonary infection localized to the left upper lobe. Her course demonstrates the di culty in treating MAC pulmonary infection despite long courses of multiple anti-mycobacterial agents. Though, she had negative sputum cultures after initiation of medical therapy, the persistence of symptoms and cavitation raised concerns about a sustained microbiological clearance, with the cavitary lesion acting a reservoir for disease. In particular, her case highlights that patients with well localized disease with few comorbidities may bene t from surgical resection compared to medical therapy alone.

Conclusions
Though surgery for pulmonary Mycobacterium avium complex disease is rarely performed, it appears as safe and at least as effective as medicaltherapy alone in our institutional experience identifying patient localized infections amenable to surgical resection. Future studies should focus on identifying whether surgical resection may offer a mortality bene t or improved microbiologic response rates by analyzing surgical outcomes at high volume surgical centers or utilizing larger registries to better power studies for which we provide deidenti ed data.

Consent to publish
No identifying patient information was included in this case series. This studied was reviewed by Partners Health System Institutional Review Board waiving need to obtain individual patient consent.

Availability of data and materials
We provide a de-identi ed patient database that can be included in future meta-analyses to better assesses treatment outcomes at https://github.com/farhat-lab/Non-tuberculous-Mycobacteria-Chart-Review/blob/master/Deidenti ed%20MAC%20Patient%20Data.csv