Methods

Apollo Cancer Institute is a 300-bed hospital that predominantly provides oncology care and belongs to a group of hospitals. Blood cultures were incubated in BacT/ALERT bottles (Biomerieux, Marcy-l‘Étoile, France) and identification was done by VITEK 2 compact analyzer (Biomerieux). We cultured 50 environmental samples, 24 intravenous fluid samples, 24 medications, and 12 refrigerated and reconstituted antibiotic preparations. ^{Reference Baul, De, Mandal, Roy, Dolai and Chakrabarti1} We collected surveillance blood cultures from all inpatients and outpatients with central lines or ports who had received cancer chemotherapy. Drug samples were cultured in 2 hospitals simultaneously. In our hospital microbiological identification was done by VITEK 2 whereas in the other hospital of our group it was done by MALDI-TOF (VITEK MS, Biomerieux). For whole-genome sequencing (WGS), we sent isolates to the Christian Medical College, Vellore, where WGS was conducted on an Illumina HiSeq 2500 platform (Illumina, San Diego, CA). Sequence types were identified using the MLST finder (https://cge.food.dtu.dk/services/MLST/). Resistance genes were identified using AMRfinderplus (https://github.com/ncbi/amr).

Between August 17 and December 24, 2021, 56 patients were identified with B. cenocepacia bacteremia (Fig. 1). The index patient in August 2021 was a child with chronic granulomatous disease who developed B. cenocepacia bacteremia while receiving treatment in an outside hospital and continued to grow B. cenocepacia from blood cultures upon transfer to our hospital. The isolate from the other hospital was resistant to ceftazidime, whereas the patient’s subsequent B. cenocepacia isolate in our hospital as well as all B. cenocepacia isolates from patients in this outbreak were susceptible to ceftazidime.

All but 1 patient had central lines or chemotherapy ports and had recently received chemotherapy, spanning from 2 days to a few weeks prior to developing bacteremia.

Among environmental samples, the gel inside the sealed ice packs (used to transport reconstituted chemotherapy drugs from the mixing cabinet to the bedside) grew B. cenocepacia, but the isolates’ antibiotic susceptibility pattern (antibiogram) was different. All ice packs were replaced with brand-new ones.

Another hospital in our group had reported an outbreak of B. cepacia bacteremia a few months earlier, in which the source was identified as ultrasound jelly. ^{Reference Yamunadevi, Ramasubramanian, Senthur Nambi, Samundeewari and Ramakrishnan2} We therefore cultured several opened and unopened bottles of ultrasound and electrocardiogram jelly. A few of these bottles grew B. cenocepacia, but these isolates also had a different antibiogram from the patient isolates in the current outbreak. Although all hospitals in our group had called back the entire stock of contaminated ultrasound jelly, we discovered that a few bottles were left behind in the cabinets at the end-user level and ensured a proper recall by taking the appropriate steps.

Chart review revealed that only oncology patients had B. cenocepacia infections and that patients belonging to only certain oncologists had B. cenocepacia bacteremia. Analysis of the drug, fluid, and device utilization patterns of all the oncologists in the hospital established that 5 oncologists were using palonosetron as an antiemetic, while others, who had no infected patients, used ondansetron. We cultured several vials of palonosetron. Of the 5 vials of a particular brand of palonosetron, 4 grew B. cenocepacia. The antibiogram of all isolates was similar to the clinical isolates. Furthermore, 5 additional vials of palonsetron were cultured in a separate laboratory of another hospital in our group; 4 of these 5 vials grew B. cenocepacia with the same antibiogram. However, vials of palonosetron belonging to a different brand used in our hospital during the same period were culture negative. The culprit batch was manufactured in July 2021, coinciding with the onset of the outbreak in August 2021. The batch was immediately withdrawn from the pharmacy.

Moreover, 4 clinical isolates and 4 drug isolates were sent for WGS analysis. All 8 isolates of B. cenocepacia belonged to ST217. We examined the SNPs in their core genome. The SNP distance comparison matrix revealed differences in only 40 SNPs, suggesting a common origin and clonality. The SNP matrices in B1 and B3 were similar. A phylogenetic tree based on SNPs showed all 8 isolates belongs to 1 phylogenetic cluster, which indicates clonality demonstrating the epidemiological link (Fig. 2).

Figure 1.

Weekly incidence of new cases.

Figure 2.

Maximum likelihood phylogenetic of Burkholderia cepacia complex. Outbreak isolates (marked as red box in phylogenetic tree) are very close to the Indian isolates that have been reported previously.