Fatal Pseudomonas aeruginosa pneumonia in a previously healthy woman was most likely associated with a contaminated hot tub

On 14 February 2010 a previously healthy 49-year-old woman presented to the emergency room of a German hospital, 2 days after returning from a spa holiday in a wellness hotel in Austria. She complained of severe chest pain with productive cough, supposedly without fever or chills. With the exception of smoking (at least 25 pack-years), there were no underlying risk factors. Initial physical examination revealed normal body temperature, blood pressure 60/40 mmHg, pulse 120/min and a respiratory frequency of 20/min. Laboratory examinations demonstrated elevated C-reactive protein (38.1 mg/dl, normal <0.5 mg/dl) and a white blood cell count of 3.94/nl (normal 4–11/nl). Antibiotic treatment was initiated with piperacillin/sulbactam [piperacillin 4 g three times daily plus sulbactam 1 g twice daily, intravenous (IV)] and moxifloxacin (400 mg once daily, IV). Blood cultures and respiratory specimens were taken and grew P. aeruginosa susceptible to piperacillin, piperacillin/sulbactam, ciprofloxacin and gentamicin when tested according to Clinical and Laboratory Standards Institute criteria [3]. PCR tests for influenza A virus, influenza B virus, cytomegalovirus, Epstein–Barr virus, herpes simplex virus and adenovirus were negative. A chest X-ray and computed tomography scan showed subtotal infiltration of the left lung. Due to rising partial pressure of carbon dioxide in the blood (pCO₂) and respiratory acidosis the patient was transferred to a university hospital several hours after initial admission and extra-corporal membrane oxygenation (ECMO) was initiated. Blood cultures and respiratory specimens again grew P. aeruginosa. On February 22, day 9 of hospitalization, the patient died of septic multiorgan failure. Two lung specimens gained at autopsy grew P. aeruginosa.

The patient fell ill 2 days after returning from a 6-day vacation at an Austrian wellness resort. According to her companion, she had repeatedly used the apartment’s hot tub. On February 26, 1 day after receiving a request issued by a German court to investigate a possible causative role of the spa, nine water samples were taken by health officers from the hotel facilities used by the patient: (1) in the apartment, from the shower, the sink and the hot tub (usually with a working water temperature of 35°C); (2) in the hotel’s wellness area, from a sauna basin, two showers, a shower in a rest room (toilet area), a small outdoor pond (created for swimming) and a knee-deep pine bath. In the hot tub sample (water temperature at the time of sampling: 18°C) 37,000 colony-forming units (CFU) of P. aeruginosa/100 ml were detected. Another strain of P. aeruginosa was isolated at 5 CFU/100 ml from the shower in the patient’s room. All other samples were negative for P. aeruginosa.

At the German reference centre for Gram-negative pathogens (Bochum, Germany), the eight isolates were typed by pulsed-field gel electrophoresis (PFGE), as described by Johnson et al. [4]. The patient’s two post-mortem lung isolates showed a PFGE band pattern indistinguishable from each other and were considered genetically indistinguishable according to the criteria by Tenover et al. [5]. From the six water isolates, four isolates (three from the hot tub and one from the shower) yielded PFGE patterns clearly distinguishable from those of the patient’s isolates and considered to be unrelated according to the criteria by Tenover et al. [5] The two remaining isolates (both from the tub) showed PFGE patterns highly similar to the lung isolates (Fig. 1). One water isolate differed from the patient’s isolates by an additional band explainable by a single insertion; the other differed by two additional bands, which could be explained either by a single insertion containing a SpeI restriction site or by two insertions. Both isolates can be regarded as being closely related based on the criteria of Tenover et al. [5]. Similar PFGE band patterns were not found among 60 P. aeruginosa isolates (49 clinical isolates, 11 environmental isolates) investigated in the laboratory performing the PFGE.

In order to assess the epidemiological relevance of the minor differences observed among the PFGE patterns of P. aeruginosa from the two patient isolates and these two tub isolates, we prospectively collected 78 clinical isolates (from June 20 to July 10, 2010, provided by 11 hospital laboratories from 5 different provinces) to compare with the eight case-related isolates (two lung isolates, six water isolates) and two epidemiologically unrelated water isolates. The clinical isolates originated from wound swabs (n = 34), tracheal secretions (n = 14), ear swabs (n = 8), urine (n = 7), blood cultures (n = 4), sputum (n = 4), eye swabs (n = 2), tongue swab (n = 1), stool specimen (n = 1), drain (n = 1) and a percutaneous endoscopic gastrostomy (PEG) tube (n = 1); for one clinical isolate no specific information was available (n = 1).

The Austrian Agency for Health and Food Safety typed the 88 isolates by PFGE pattern analysis, as described above, by amplified fragment length polymorphism (AFLP) analysis and by multi-locus sequence typing (MLST). Genomic bacterial DNA (gDNA) for the AFLP and MLST analysis was extracted with the Quick Extract Bacterial DNA Extraction kit according to the manufacturer’s instructions (Epicentre Biotechnologies, Madison, WI). The AFLP analysis was performed using an AFLP Core Reagent kit (Life Technologies, Carlsbad, CA) for gDNA restriction, adaptor ligation and prePCR according to the manufacturer’s protocol [6]. MLST analysis was performed on all 88 isolates as described by Mansfeld et al. [7]. The sequences obtained were submitted to the P. aeruginosa PubMLST database (http://​pubmlst.​org/​paeruginosa) [8]. Novel alleles and sequence types (STs) were submitted for allele and ST and clonal complex (CC) designations.

The 88 P. aeruginosa isolates were attributable to 56 sequence types as determined by MLST, to 80 PFGE types and to 74 AFLP types. Nine new alleles and 22 new STs, previously not available in the P. aeruginosa MLST database, were found [8]. Twelve STs comprised at least two different isolates. Twenty-two STs were assignable to ST groups within the MLST database using eBURST v3 (http://​eburst.​mlst.​net) cluster analysis. The remaining 34 STs represented singletons.

The two lung isolates #74 and #75, the two hot tub water isolates #76 and #78 and an epidemiologically unrelated clinical isolate derived from tracheal secretions of a patient from the province Upper Austria (#27) belonged to ST-313 and could clearly be differentiated from the remaining 83 isolates (Fig. 2).

The AFLP pattern analysis of these five isolates revealed three different patterns (Fig. 2). One band of the epidemiologically unrelated isolate #27 had a lower molecular weight than the respective band of the two lung isolates #74 and #75 and the water isolate #78. The water isolate #76 lacked this band completely. The AFLP patterns of these five isolates were clearly different from those of the remaining 83 isolates, similar to the PFGE patterns being clearly different from those of the 83 isolates. PFGE pattern analysis revealed three different patterns (Fig. 2). The epidemiologically unrelated isolate #27 lacked one band present in the two lung isolates. One band of the water isolate #76 had a higher molecular weight than that of the respective band of the two lung isolates. The PFGE pattern of water isolate #78 was indistinguishable from that of the epidemiologically unrelated isolate #27. According to the criteria of Tenover et al. [5], the lung isolates and the water isolates from the tub can therefore to be considered as genetically closely related and as standing in an epidemiological context.