MYCOBACTERIUM CHIMAERA INFECTIONS AFTER CARDIOVASCULAR SURGERY: LESSONS FROM A GLOBAL OUTBREAK

Abstract

A global outbreak of invasive Mycobacterium chimaera infections has been associated with exposure to certain heater-cooler devices (HCDs) used during cardiopulmonary bypass. Outbreak investigations have shown that these HCDs harbor M. chimaera in water circuits and generate bio-aerosols in the operating room, leading to airborne transmission to patients during surgery. Whole genome sequencing data support a common-source outbreak originating at an HCD manufacturing facility. Most clinical infections are associated with implanted devices, diagnosis is often delayed, and treatment requires device removal and prolonged antibiotic therapy. Because it is nearly impossible to eradicate M. chimaera from HCDs using existing disinfection approaches, strict separation of HCD exhaust from operating room air is necessary to prevent patient exposure. Lessons learned from this outbreak include: 1) medical device risks are difficult to predict, requiring improved expert review before approval, and 2) advances in genomics provide powerful tools for outbreak investigation and public health surveillance.

THE OUTBREAK

Beginning in 2013, a small number of patients at the University of Iowa Hospitals and Clinics presented with a similar history of fevers, night sweats, and weight loss. Physical examination in each case revealed hepatosplenomegaly, and lab findings included pancytopenia and elevated transaminases. When biopsies were performed (of sites including liver, bone, and bone marrow) noncaseating granulomas were seen, but all organism stains were negative. After extensive evaluation for infectious and noninfectious etiologies was unrevealing, some patients were started on steroids for presumed sarcoidosis or other autoinflammatory conditions.

Similar cases were occurring simultaneously in Switzerland, the United Kingdom, and in two health systems in Pennsylvania (1-3). In every case, there was history of cardiac surgery that included cardiopulmonary bypass and involved the placement of a device (prosthetic valve or graft, or left ventricular assist device).

Eventually, these patients were all diagnosed with invasive Mycobacterium chimaera infection after mycobacterial cultures of blood, bone marrow, or another sterile body site were obtained. M. chimaera is one of the species in the M. avium complex, and is a ubiquitous environmental organism that generally causes disease only in those with underlying airway disease (e.g., bronchiectasis), or in individuals who are profoundly immunocompromised (e.g., those with AIDS with CD4 counts of < 50 cells/mm³) (4,5). However, in these cases the infections involved prosthetic devices, from which the organism disseminated to involve other organ systems. As more such cases were reported from multiple states and countries (now with more than 150 confirmed cases worldwide), it became clear that there was an ongoing global outbreak of invasive, device-associated infection with M. chimaera.

The clinical aspects of this outbreak are still to be fully elucidated, but case reports thus far describe several common features (6-8). First, the diagnosis of M. chimaera infection is often delayed, sometimes for several years after cardiac surgery. Delayed diagnosis is due both to the insidious onset of infection (with a mean time from surgery to symptom onset of approximately 18 months), and to the difficulty in isolating M. chimaera in culture. Not only do special cultures for mycobacteria have to be performed, but the organism can takes weeks to grow. Second, the crude mortality of the outbreak has been very high, close to 50%. The affected patients often have multiple comorbidities, and by the time the diagnosis is established they may be extremely ill. Finally, the treatment of infection is extremely challenging, requiring multiple drug therapy (usually at least three drugs, including a macrolide, a rifamycin, and ethambutol) and surgical removal of any involved devices. Drug interactions and toxicities, and the substantial risks of device removal, complicate therapy and add to the mortality risk.

OUTBREAK INVESTIGATION

The first clues to the cause of the M. chimaera outbreak were reported in 2015 by Sax et al. (3). After six patients with a history of cardiac surgery developed device-associated M. chimaera infection at their hospital, they performed a careful investigation for possible environmental sources. During this investigation, they found that M. chimaera was present in the water from the reservoir of the heater-cooler devices (HCDs) used during cardiopulmonary bypass. Moreover, they found that the air samples taken in the operating room (OR) also grew M. chimaera, but only when the HCDs were in operation (3). Furthermore, a Centers for Disease Control (CDC)–led investigation of the M. chimaera outbreak in Pennsylvania confirmed that exposure to HCDs during cardiac surgery (particularly prolonged exposure) was the most important risk factor for developing invasive M. chimaera infection (2).

HCDs are used to regulate temperature during cardiopulmonary bypass, and the HCD implicated in the outbreak cases (LivaNova 3T Heater-Cooler System, LivaNova, London, UK) has three water circuits: one to a patient heating/cooling blanket, one to the oxygenator of the bypass pump, and one to the cardioplegia solution (9). The water in these circuits returns to a water reservoir within the HCD. The water in the water circuits never directly contacts the patient or the patients' blood, and the circuits are intended to be watertight.

However, careful studies have shown that aerosols from inside the water reservoir of the implicated HCD could escape and be exhausted from the back of the HCD via a ventilation fan (1). Smoke testing in a simulated OR then showed clearly that the air exhausted from the HCD could reach the patient if the HCD was oriented with the fan facing toward the operative field (10). Meanwhile, water cultures from Stöckert 3T HCDs in multiple centers in different states and countries were found to contain M. chimaera (8).

While the above findings established a plausible route of infection (exposure of the patient and/or the implanted devices to M. chimaera–laden bioaerosols during surgery), the source of the outbreak organism was still in question. Although contamination of the HCDs at the individual hospitals was possible, the fact that all infections were due to a single unusual species of mycobacteria suggested a common source. To investigate the genetic relatedness of the outbreak isolates, we collaborated with investigators at National Jewish Health in Denver, Colorado, and the CDC to do whole-genome sequencing (WGS) on M. chimaera isolates from three case patients and six HCDs from Iowa. Isolates from centers in six other states, Italy, and Switzerland, and previously published sequences from additional US states, Australia, Denmark, and New Zealand were also included in the analysis. Among the more than 100 isolates, almost all the outbreak-related strains clustered together, with a mean pairwise difference in single nucleotide polymorphisms (SNPs) of 4 (range: 0 to 23 SNPs) (11), whereas the epidemiologically unrelated isolates of M. chimaera differed from the outbreak isolates by a mean of more than 500 SNPs. Other published reports confirm these findings, suggesting a common source (or “point source”) outbreak (12-15).

Haller et al. (16) established that this point source was likely to be at the manufacturing plant by culturing M. chimaera from factory-new HCDs as well as from a water source in the pump assembly area of the manufacturing plant. A factory water isolate was also part of the outbreak cluster when included in the WGS analysis (14). The conclusion, as outlined by van Ingen et al. (14), is that “it is most probable that LivaNova HCDs represent the common source of infection and that contamination of most of these HCDs occurred during production….”

OUTBREAK RESPONSE

The discovery of a global outbreak of life-threatening M. chimaera infections associated with a widely used medical device posed a major outbreak response challenge to hospitals that performed cardiac surgery, as well as to public health and regulatory agencies (e.g., the US Food and Drug Administration [FDA]). The two immediate priorities were: 1) raising awareness to improve case finding; and 2) mitigating the exposure risk by advising hospitals regarding disinfection and use of their HCDs. The CDC and other state and national public health agencies issued alerts to clinicians and hospitals to guide them in patient notification and case-finding efforts (17), and the FDA and the manufacturer issued updated recommendations for disinfection and use of the involved HCD (18).

Case-finding challenges include the fact that symptom onset is delayed from time of surgery, many patients have their follow-up care at a substantial distance from where their cardiac surgery was performed, and there is no screening test to assess asymptomatic patients for exposure or infection (19). As a result, many patients may yet be undiagnosed, and some have undoubtedly died before diagnosis.

Risk mitigation is also challenging. Because the Stöckert 3T has the largest share of the HCD market (60% to 80% worldwide), it is not possible to recall all 3T units without disrupting the availability of lifesaving cardiac surgery. Unfortunately, several investigators have also shown that M. chimaera is virtually ineradicable from HCDs once it has colonized the water circuit, despite more intensive disinfection approaches (20-22). The lipid-rich cell wall of organisms within the M. avium complex and the high concentration of such organisms within biofilms render them highly resistant to standard disinfectants as well as making them highly amenable to aerosolization from water circuits (23,24). For similar species within the M. avium complex, the concentration of organisms within water bubbles aerosolized from water columns is believed to be up to 10,000-fold higher than the concentration within the water source (25).

Thus, the only effective risk elimination strategy is likely to be complete separation of the HCD exhaust air from OR air. We immediately removed our HCDs from the OR by building a pass-through in the OR walls. This allowed the tubing to be connected to the HCD and the HCD to be operated remotely by the perfusionist from inside the OR (19). Others have used different approaches to this problem, including building encasement devices designed to direct HCD exhaust through a HEPA filter or directly out of the OR (22).

LESSONS LEARNED

A major lesson from this outbreak is that the infection risk from a medical device can go unrecognized for long periods. HCDs have been used during cardiopulmonary bypass for decades, and yet the risk of bioaerosol formation has not been widely recognized until this outbreak. One group published a prescient article raising this concern, but it was not published in a journal with wide reach and was thus not heeded (26). Other HCD-associated outbreaks have now been recognized (27), and it is not known how many sporadic cases of mediastinitis, prosthetic valve endocarditis, or vascular graft infection may have resulted from HCD bioaerosol exposure. Meanwhile, other advanced medical devices have also been recognized as important infection risks well after their introduction, such as the duodenoscopes linked to large outbreaks of multiple-drug resistant organisms due to inability to effectively disinfect the intricate mechanisms within the device (28). This indicates a need for improved review of the safety of medical devices both by regulatory agencies and by individual hospital infection prevention programs. Certain principles may be applied here, as well; e.g., any device that includes a water source and a fan should clearly undergo careful review, or perhaps not be allowed in critical areas such as ORs.

Another lesson from this outbreak is the power of advanced molecular diagnostics in outbreak recognition, investigation, and response. WGS allows for sequences to be quickly compared across the globe to facilitate the investigation of transmission pathways and potential point sources, as occurred in this outbreak (29). Public health surveillance is increasingly moving forward with next-generation sequencing, which will increase our ability to detect and respond to outbreaks earlier, and to detect and monitor new pathogens with important resistance or virulence markers.

FUTURE DIRECTIONS

Future priorities in response to this global outbreak include a need to redesign heater-cooler technology for cardiopulmonary bypass. HCDs must be designed in such a way as not to be able to produce bioaerosols. There may also be options for heating and cooling during bypass that do not require HCDs at all, as recently described by Matte et al. (30).

There also must be improved infection control scrutiny of medical devices, as described above. Regulatory agencies such as the FDA should include reviews by individuals with expertise in hospital infection prevention, and each institution should also review all new devices at time of introduction to help detect any potential risks for infection transmission.

Improved availability of advanced molecular diagnostics, including WGS for outbreak response, must also be a priority. The CDC's PulseNet program is already transitioning from older typing methods to WGS for foodborne pathogen surveillance (31), but these technologies must also be more readily available to healthcare facilities investigating nosocomial threats. The CDC's Advanced Molecular Detection initiative to fund state and regional public health laboratories to perform such testing on target pathogens is an excellent first step (32).

Finally, the many challenges in diagnosis and treatment of patients involved in this global outbreak call for a patient registry to be established. Such a registry could help inform patient care, and also provide a better understanding of the outcomes of device-associated M. chimaera infection, including how outcomes differ based on treatment approaches (both antimicrobial and surgical). Such a registry has recently been funded by the CDC and will be managed at the University of Iowa.

DISCUSSION

Hook, Birmingham: Very nice, Dan. I was wondering if you wanted to comment about the challenge of biofilms in our devices as sources of infection and whether there are good ways to address them?

Diekema, Iowa City: I think that's a great point. If you look at healthcare-associated infections, it's estimated that 80% to 85% or more are associated with devices. Every one of those, in general, involves a biofilm. One of the real holy grails is in materials engineering to design surfaces that resist biofilm formation or to design agents that can actually breakdown biofilm. Without doing that, it becomes virtually impossible to eradicate those infections. This isn't just device-associated. We have cystic fibrosis researchers here who know that a lot of infections in the human body that aren't device-associated are also closely biofilm-associated or it plays a major role in the pathogenesis. In all of the things we do in infectious diseases, measuring minimum inhibitory concentrations and determining which antibiotics might work, are all done in a planktonic model that doesn't take into account, in any way, the presence of organisms within a complex matrix that a biofilm is. So, I don't have any answers but I think that's one of the future directions.