By Carol A. Kemper, MD, FACP

Clinical Associate Professor of Medicine, Stanford University, Division of Infectious Diseases, Santa Clara Valley Medical Center

Dr. Kemper reports no financial relationships relevant to this field of study.

Fatal ESBL Infection from Fecal Microbiota Transplant

SOURCES: DeFilipp Z, Bloom PP, Torres Soto M, et al. Drug-resistant E. coli bacteremia transmitted by fecal microbiota transplant. N Engl J Med 2019;381:2043-2050.

Blaser MJ. Fecal microbiota transplantation for dysbiosis – predictable risks. N Engl J Med 2019;381:2064-2066.

Fecal microbiota transplant (FMT) has become an accepted treatment for severe and refractory Clostridioides difficile infection (CDI). There has also been significant interest in the use of FMT for replacement of fecal flora in patients with intestinal dysbiosis, as a way to prevent pathogenic organisms from flourishing in such compromised patients. These authors described two patients who received FMT in the context of clinical studies for the treatment of non-CDI-related conditions, both of whom developed extended spectrum beta-lactamase (ESBL)-containing Escherichia coli with a donor strain. One patient died of severe sepsis.

The first patient had severe hepatic encephalopathy and received 15 FMT capsules five times over a three-week period, in addition to rifaximin. Seventeen days after his final FMT dose, he developed cough and pneumonia, and blood cultures grew ESBL-containing E. coli. He recovered with carbapenem treatment, and peritoneal fluid and stool samples (on selective media) were negative for ESBL. The second patient had myelodysplasia and was enrolled in a Phase II trial to administer FMT oral capsules before and after allogeneic stem cell transplant. He received 15 FMT capsules on the third and fourth days before his transplant. Eight days after his last FMT dose, and five days after his stem-cell infusion, he developed sepsis and quickly died. Blood cultures grew ESBL-containing E. coli. Frozen stool samples obtained from both patients before their FMT treatment were negative for ESBL.

The FMT product given to both patients was derived from the same donor. A sample of this product proved to contain ESBL E. coli and whole-genome sequencing demonstrated a match between this isolate and the two clinical isolates ( 1 single nucleotide polymorphism [SNP] difference). Twenty-two patients also received FMT from this donor in the context of these two clinical trials, and another 16 for the treatment of CDI. None of the stool samples from patients in the clinical trial were positive for ESBL; however, five of 12 patients treated for CDI also were positive for ESBL E. coli. Stored capsules from 32 other FMT samples from 10 other donors also were cultured on exclusion media, none of which were positive for ESBL. This suggests the frequency of ESBL-positive stool specimens found in patients with CDI receiving this donor’s stool was significant.

Extensive systematic screening of FMT donors is performed according to Food and Drug Administration (FDA) requirements. Donors are healthy, asymptomatic, and between the ages of 18 and 50 years; all donations are frozen and unused for four weeks in case of subsequent illness. Donated stool is processed, concentrated, placed in capsules, and stored frozen at -80°C. The frozen capsules these patients received had been stored for nine months in accordance with approved procedures. FDA regulatory requirements were amended in January 2019 to include screening for ESBL-containing organisms, after this product was manufactured.

Screening for ESBL-containing organisms is increasingly useful for hospital epidemiology, especially as we battle hospital-acquired infections. Our facility has been performing perirectal swabs to rule out ESBL and carbapenem-resistant Enterobacteriaceae (CRE) in high-risk admissions (e.g., skilled-nursing facility [SNF] patients and travelers from other countries) for several years (using exclusion media), although the process is disliked by patients and it is rather expensive as a surveillance tool. In 2017, the frequency of ESBL colonization in our SNF population was ~17%. We also see an increased frequency of ESBL and CRE colonization in patients from India and Southeast Asia, and in travelers to developing countries. Based on the cost (and the lack of reimbursement for surveillance testing from Medicare), we have had to abandon ESBL screening and now just perform CRE screening in select high-risk admissions. But the problem has always been the reduced sensitivity of the perirectal screening, even using exclusion media. Several patients with clinical ESBL infection of urine and blood initially had negative stool or perirectal screening, only to have positive results weeks or months later. Only after the selective pressure of broad-spectrum antibiotics may the fecal ESBL or CRE appear in some patients. Undoubtedly, the two patients discussed earlier had intestinal colonization with ESBL from their ingested FMT with translocation to the bloodstream — not detected by stool cultures.

The point is, even if the FDA has adopted measures to screen donated stool for resistant gram-negative organisms, the sensitivity of this surveillance is not known — and may not be adequate to the task. At a minimum, FMT probably should not be used in immune-compromised patients — but I am not sure we want to be giving ESBL or other resistant bacteria to anyone (there is no good evidence that, once colonized, you can become “uncolonized”). Perhaps we should consider using donor stool from a spouse. As I have told patients, if you have exchanged other bodily fluids for years, and raised children, and your donor stool is free from any obvious pathogens, maybe it is better to use your sexual partner’s stool (if you are lucky to have one) — or an appropriately screened close family/household member.

Second Joint Infection When One Prosthetic Gets Infected?

SOURCE: Wourthuyzen-Bakker M, Sebillotte M, Arviewx C, et al. How to handle concomitant asymptomatic prosthetic joints during an episode of hematogenous PJI, a multicenter analysis. Clin Infect Dis 2020; Aug. 18;ciaa1222. [Online ahead of print].

Prosthetic joint infection (PJI) as a consequence of bloodstream infection is a devastating event, frequently resulting in weeks of antibacterial therapy, surgical debridement, and, often, explantation of the joint. The overall risk of PJI is ~0.07% per year of life of the prosthesis, with knee joints at the highest risk. Since many modern patients may have more than one PJ, should a PJ become infected with complicating bacteremia, what is the risk to their other PJs?

These authors examined the risk of infection to a concomitant PJ as a consequence of bacteremia with an acute hematogenous PJI. A total of 91 patients with hematogenous PJI who also had 108 other PJs were assessed (including 56 knees, 44 hips, three shoulders, and five others). Infection of the first PJ was caused by Staphylococcus aureus (43%), streptococci (26%), and gram negatives (18%), with bacteremia in all cases. Thirteen patients experienced symptoms of possible concurrent PJI in an additional joint, including seven patients with acute joint pain (54%), two patients with acute on chronic pain (15%), and four patients with chronic pain (31%). All 13 had clinical signs of infection on examination. However, only 10 of these additional joints proved to be infected, and symptoms in the other three were attributed to mechanical reasons. One additional PJI became apparent in subsequent weeks, giving a total of 11/108 (10.2%) concomitant joint infection in these patients. In patients with both prosthetic hips and knees, the knee joint was more likely to be infected than the hip (78% vs. 22%). None of the other PJs were infected. Interestingly, infected joints were younger than uninfected joints (4.5 vs. 6.7 years; P = 0.04). The risk that a concurrent joint infection occurred as the result of a first hematogenous joint infection with S. aureus was 2.7% to 5.6%, depending on the joint.

This work complements the study summarized by Stan Deresinski in the January 2020 issue of Infectious Disease Alert, wherein the risk of PJI during an episode of bacteremia largely depends on the organism.1 Honkanen and colleagues observed that PJI occurred with 46/643 (7%) episodes of bacteremia in 45/542 (8%) patients (one patient had two episodes). The median time interval from the first positive blood culture to identification of PJI ranged from 0-522 days, with a median interval of two days. While 1.3% of E. coli and 1.4% of other gram-negative bacteremias resulted in PJI, 20% of bloodstream infections from S. aureus caused PJI.

The present study extends this work to examine the risk to a concurrent joint during an episode of PJI and complicating bacteremia. Of course, often it is difficult to appreciate whether the bacteremia was first and the joint infection second, or whether the joint infection gave rise to the bacteremia. Presumably, many of these infections are hematogenous in origin, unless related to the original surgery. The risk of prosthetic joint infection is significantly higher with S. aureus bacteremia, which generally causes higher grade endovascular infection. Nonetheless, it was surprising to see that S. aureus infection in one PJ with bacteremia had a 2.7% to 5.6% chance of involving a second prosthetic joint.


  1. Honkanen M, Jämsen E, Karppelin M, et al. Periprosthetic joint infections as a consequence of bacteremia. Open Forum Infect Dis 2019;6:ofz218.