Disseminated Ureaplasma infection as a cause of fatal hyperammonemia in humans

Index case

A 44-year-old man with pulmonary sarcoidosis underwent bilateral lung transplantation at Northwestern Memorial Hospital (Chicago, IL) with induction immunosuppression consisting of methylprednisolone (500 mg) and basiliximab (20 mg). Primary graft dysfunction necessitated 2 days of extracorporeal membrane oxygenation. Maintenance immunosuppression included tacrolimus (target trough level, 8 to 12 ng/ml), mycophenolate mofetil (1000 mg twice daily), and prednisone (0.5 mg/kg). He received antimicrobial prophylaxis with trimethoprim-sulfamethoxazole, valganciclovir, and voriconazole. A 7-day course of vancomycin and cefepime was administered to treat Staphylococcus aureus, Stomatococcus species, and viridans group streptococci isolated from bronchoalveolar lavage (BAL) fluid. Seven days after transplant, he developed acute mental status changes. Routine blood, urine, and BAL fluid cultures were negative, and head computed tomography was normal. Serum ammonia, which had been normal on postoperative day 4, was 61 μM on day 5 (laboratory reference range, 0 to 53 μM) and 107 μM on day 7 (Fig. 1). Testing to exclude known abnormalities in urea cycle metabolism was negative (table S1). Venovenous hemodialysis for ammonia removal was initiated; tacrolimus and protein in the tube feedings were temporarily discontinued; and lactulose, rifaximin, sodium phenylacetate/benzoate, and L-carnitine were administered according to an established protocol (table S2) (2). Despite these therapies, the serum ammonia rose to 291 μM on day 10. On the basis of a reported association between Mycoplasma hominis infection and hyperammonemia (6), the patient received azithromycin (500 mg daily for 5 days), beginning 10 days after transplant. Serum ammonia concentration began to decrease the day after azithromycin initiation and was normal on day 15 (Fig. 1A), and this normalization was accompanied by recovery of his mental status. Recurrent hyperammonemia developed on postoperative day 31 (Fig. 1B) and worsened despite reinstitution of the therapies above, including parenteral azithromycin. On postoperative day 40, he died from a neurologic event, and autopsy revealed cerebral edema and pulmonary consolidation.

Molecular diagnostics

We undertook a careful microbiologic examination of specimens collected during the patient’s hospitalization (Table 1, microbiologic results for non-Ureaplasma organisms are presented in table S3). Examination of plasma by polymerase chain reaction (PCR) first detected U. urealyticum 9 days after transplantation, 1 day before the initiation of azithromycin therapy. This specimen was negative for M. hominis by PCR and culture. Examination of plasma for U. urealyticum by PCR was positive 1 day before starting azithromycin and negative immediately after treatment (Fig. 1A and Table 1). Samples of bladder, liver, spleen, and lung tissue obtained less than 6 hours postmortem tested positive for U. urealyticum and negative for U. parvum, M. hominis, and Mycoplasma pneumoniae by PCR. U. urealyticum was cultured from the lung at autopsy and was resistant to azithromycin and erythromycin [minimum inhibitory concentration (MIC) values, 256 μg/ml] but susceptible to levofloxacin (MIC, 1 μg/ml), tetracycline (MIC, 0.5 μg/ml), and doxycycline (MIC, 0.25 μg/ml). Sequencing of this U. urealyticum isolate identified a A2058G mutation in the 23S ribosomal RNA (rRNA) gene operon 2 (Escherichia coli numbering system) associated with macrolide resistance (7, 8). No other mutations were detected in this gene or the genes for proteins L4 or L22 of the ribosomal unit. Native lung tissue from the recipient was negative for these organisms.

Additional patients with hyperammonemia and control cohort

We then performed microbiologic analysis of other patients who died of HS at three different centers, as well as 20 lung transplant recipient controls with normal ammonia concentrations (6). The first patient was a 59-year-old man who underwent single-lung transplantation at Washington University (St. Louis, MO) for emphysema, developed HS refractory to the therapies described above 10 days after transplant, and died from cerebral edema with herniation. Freshly frozen, paraffin-embedded lung tissue collected at autopsy tested positive for U. urealyticum but negative for U. parvum and M. hominis by PCR. The second patient was a 64-year-old man who had undergone bilateral sequential lung transplantation at Mayo Clinic (Rochester, MN) for pulmonary fibrosis and died 16 days after transplantation from cerebral edema. Liver function was normal, and liver biopsy revealed normal architecture; glutamine synthetase levels were normal, as described in a report previously published on this patient (6). This patient was shown to have systemic M. hominis infection (6). We performed PCR and found that residual BAL fluid and blood collected shortly before his death were positive for both M. hominis and U. parvum. The third patient was a 59-year-old man who underwent bilateral lung transplant for idiopathic pulmonary fibrosis at Loyola Medical Center (Chicago, IL). On postoperative day 12, he developed mental status changes, and his ammonia concentration was 549 μM. Despite treatment similar to that outlined in table S2, the ammonia concentration increased to greater than 1700 μM, and the patient expired. Freshly frozen, paraffin-embedded lung tissue collected at autopsy tested positive for U. urealyticum but negative for U. parvum and M. hominis by PCR. We obtained the donor BAL collected before lung harvest, as well as the recipient’s BAL collected immediately after implantation, and found that both tested positive for U. urealyticum but negative for U. parvum and M. hominis. The recipient’s native lung tissue was negative for all three organisms.

We then tested BAL fluid and serum collected from 20 lung transplant recipients who had documented normal ammonia concentrations 1 to 7 days after transplant; all were PCR-negative for U. urealyticum, U. parvum, and M. hominis. Cultures could be obtained from five of these patients and were also negative for the same three organisms (Table 1).

Prospective validation

On the basis of these results, we tested for Ureaplasma organisms in blood and BAL fluid from a 52-year-old female who underwent bilateral lung transplantation at Washington University (St. Louis, MO) for interstitial lung disease complicated by refractory hyperammonemia 9 days later. Both specimens were positive for U. parvum by PCR, and BAL fluid culture was positive for U. parvum but negative for U. urealyticum or M. hominis. Two days after initiating therapy with levofloxacin and azithromycin, the ammonia concentration normalized, and her mental status returned to normal (Fig. 2). After 7 days of antimicrobial treatment, both BAL fluid and blood were negative for U. parvum, and there has been no relapse of hyperammonemia in 6 months of follow-up.

Another recent patient with HS was successfully treated, providing additional prospective validation of the relationship between Ureaplasma and HS. This was a 55-year-old female who underwent bilateral sequential lung transplant for idiopathic pulmonary fibrosis at Massachusetts General Hospital (Boston, MA). Because the lymphocyte crossmatch was positive at the time of transplant, she was treated with peri- and postoperative plasmapheresis, followed by intravenous immunoglobulin and rituximab. The remainder of her immunosuppression regimen consisted of pre-/postoperative mycophenolate and intraoperative methylprednisolone 1 g intravenously and then 125 mg every 12 hours for a total of eight doses, followed by tapering doses; basiliximab 20 mg intravenously on postoperative days 2 and 6; and tacrolimus on postoperative day 4, titrated to achieve a trough concentration of 10 to 14 ng/ml. On day 10, she demonstrated altered mental status. Computed tomography of her brain demonstrated no evidence of acute pathology, but an electroencephalogram showed bifrontal rhythmic periodic discharges concerning for nonconvulsive epileptic seizure activity. Magnetic resonance imaging revealed newly detected multiple microhemorrhages. At that time, her ammonia concentration was found to be elevated at 135 μM. Lactulose (20 g three times a day), rifaximin (550 mg two times a day), sodium benzoate (5 g two times a day), arginine (4 g three times a day), and levocarnitine (7 g/day) were initiated. Her tube feeds were adjusted to lower protein content. Hemodialysis was also initiated to facilitate removal of ammonia. Additionally, tacrolimus was discontinued on postoperative day 20 when hyperammonemia failed to resolve and replaced with cyclosporine, similar to the index patient. Although ammonia concentration transiently decreased, this response was dependent on hemodialysis. Even a temporary discontinuation of hemodialysis led to an increase in ammonia to greater than 200 μM. This pattern persisted even at day 42, with no significant improvement in mental status or consistent decrease in ammonia. At this time, her BAL and blood were tested, and both were found to be positive for U. parvum by PCR and culture. The patient received levofloxacin, and the ammonia concentration decreased to 30 to 35 μM, despite discontinuation of hemodialysis and other adjunctive methods for treating hyperammonemia. At 4 months after transplant, there has been no relapse of HS.

Culture

U. urealyticum, U. parvum, and M. hominis culture was performed at Mayo Clinic and Quest Diagnostic Nichols Institute. Specimens were cultivated on A8 agar plates (Hardy Diagnostics) in an anaerobic atmosphere at 35°C for 6 weeks. Culture plates were examined by inverted light microscopy at least weekly. Suspect colonies were confirmed by real-time PCR (21). Recovered isolates were subcultured and incubated for 48 hours. Subculture material was placed into M5 transport medium (Remel) and frozen at −20°C for transport to the University of Alabama at Birmingham (UAB) for susceptibility testing.

Antibiotic susceptibility testing

Minimal inhibitory concentration values were determined by broth microdilution in 10B broth (22), in accordance with Clinical and Laboratory Standards Institute guidelines (http://clsi.org/blog/2011/11/11/clsi-publishes-a-guideline-on-methods-for-antimicrobial-susceptibility-testing-for-human-mycoplasmas-m43/) (22).

PCR and molecular resistance profiling

PCR for U. parvum, U. urealyticum, and M. hominis was performed at Mayo Clinic as described before (21). Genomic U. urealyticum DNA was purified using the QIAamp DNA Blood Mini Kit (Qiagen). Conserved regions of ribosomal protein L4 and L22 genes and two whole 23S rRNA operons were amplified using published primers (23, 24). PCR was performed with Platinum PCR SuperMix (Invitrogen) on a Veriti 96-well thermal cycler (Applied Biosystems). PCR products were cleaned by USB ExoSAP-IT (Affymetrix) and sequenced using Sanger technology at UAB. Sequences were compared to corresponding gene sequences of 14 macrolide-susceptible American Type Culture Collection (ATCC) type strains of U. parvum (serovar 1, ATCC 27813; serovar 3, ATCC 27815; serovar 6, ATCC 27818; and serovar 14, ATCC 33697) and U. urealyticum (serovar 2, ATCC 27814; serovar 4, ATCC 27816; serovar 5, ATCC 27817; serovar 7, ATCC 27819; serovar 8, ATCC 27618; serovar 9, ATCC 33175; serovar 10, ATCC 33699; serovar 11, ATCC 33695; serovar 12, ATCC 33696; and serovar 13, ATCC 33698) using CLC DNA Workbench 7.5 (Qiagen).