July 2019: Onc and transplant ID

What predicts MDR Pseudomonas bacteremia in patients with neutropenia?  This research group out of Spain performed a retrospective study of Gram-negative bacteremia seen in neutropenic patients at their hospital over a 13-year period.  They identified a total 661 episodes of GNR bacteremia, of which 190 were caused by Pseudomonas aeruginosa (PSA), and of those 70 by MDR PSA. 

In univariate analysis, risk factors for MDR-PSA as the cause of bacteremia included prior use of an antipseudomonal cephalosporin (OR 13.7; 95% CI 6.6-28), pulmonary infection (OR 7.9; 95% CI 3.3-18.6), development of bacteremia while on ceftriaxone (OR 4.3; 95% CI 1.2-15.8), nosocomial acquisition (OR 3.5; 95% CI 1.7-7.1), hematologic malignancy (OR 3.3; 95% CI 1.2-9.5), and use of pip-tazo (OR 2.4; 95% CI 1.0-5.6).  When looking only at Pseudomonas aeruginosa bacteremia, the risk factors for MDR-PSA were nosocomial acquisition (OR 7.1), bacteremia development while on a quinolone (OR 4.9), bacteremia development while on a beta-lactam other than ertapenem (OR 4.5), prior quinolone use (OR 4), prior antipseudomonal cephalosporin use (OR 3.8), hematologic malignancy (OR 3.4), and corticosteroid use (OR 2.9).  The authors developed a clinical risk score for multidrug resistance in Pseudomonas bacteremia based on these odds ratios, though it was not validated.

Some of these risk factors are obvious – e.g.; of course you’re more likely to see a bacteremia with a ceftriaxone or cefepime-resistant organism if the bacteremia appears while the patient is taking one of those drugs.  More interesting to me was the association with quinolones.  Increasing evidence suggests that quinolone prophylaxis in neutropenic cancer patients may be associated with increased incidence of infection with MDR gram-negative organisms, and it looks like that would probably be true for MDR Pseudomonas aeruginosa as well.  I’m not necessarily arguing against quinolone prophylaxis for patients with chemotherapy-induced neutropenia – but it does suggest that when these patients do develop neutropenic fever they might benefit from more aggressive empiric therapy than their hospitalization and antibiotic exposure histories would otherwise suggest. 31295551

Immunocompromised adults do poorly when hospitalized with laboratory-confirmed influenza.  This multicenter US study, led by Emory and the CDC, examined the outcomes of immunocompromised vs immunocompetent adults hospitalized with community-acquired influenza.  Here, “immunocompromised” meant having AIDS, cancer, a history of stem cell or solid organ transplantation, immunoglobulin or complement deficiencies, asplenia, treatment with immunosuppressive agents other than steroids, or a few other rare conditions.  The researchers used an influenza surveillance system, FluSurv-NET, to capture data from hospitals and laboratories in 13 states, with a total catchment including ~9% of the US population; they defined laboratory-confirmed influenza as a positive rapid antigen, RT-PCR, immunofluorescent antibody, or viral culture test.  Patients were excluded if no information was available regarding whether they: had an ICU admission, were on mechanical ventilation, or died during their hospital stays.  The primary outcomes were all-cause in-hospital mortality, ICU admission, and length of hospital stay.

Data collected from the 2011-2012 through 2014-2015 flu seasons yielded a total 35,327 adults hospitalized with influenza, among whom 3,633 were immunocompromised.  The most common forms of immunocompromise were cancer (44%), non-steroid immunosuppressive therapy (44%), AIDS (18%), and organ transplantation (15%) (note patients could have more than one cause of immunocompromise, so the numbers don’t add to 100).  Compared to their immunocompetent peers, patients with immunocompromising conditions were younger (62 vs 70; p<0.001), more often male (53% vs 44%; p<0.001), and more likely to have received the seasonal flu vaccine (53% vs 47%; p<0.001).

Compared to their peers without immunocompromising conditions, patients with IC had similar durations of symptoms prior to presentation (4 days for both groups), and were modestly more likely to be febrile (68% vs 61%; p<0.001) and to have GI symptoms (34% vs 28%; p<0.001) - otherwise, while there were plenty of low p values on account of the big sample size, clinical presentations were not meaningfully different between groups.  Immunocompromised and immunocompetent patients received antivirals with similar frequency (87% vs 85%), and this meant osteltamivir in >99% of casers.  Immunocompromise was a risk factor for ICU admission (OR 1.1; 95% CI 1.0-1.2) and death (OR 1.3; 95% CI 1.1-1.5) in univariate analysis. In the adjusted multivariate model immunocompromise was associated with mechanical ventilation (aOR 1.2; 95% CI 1.1-1.4) and death (aOR 1.5; 95% CI 1.2-1.8), whereas immunocompromise remained associated with ICU admission only for patients over 65 years of age.  Finally, length of hospital stay was longer (4 vs 3 days; p<0.001) for the immunocompromised patients; durations of ICU stay were similar.

Bottom line: immunocompromised patients make up about 10% of hospitalizations for influenza, are more likely to present with fever and GI symptoms, and have longer stays with more ICU admissions (most clearly for older patients) and more deaths.  In their discussion, the authors emphasize the low rates of influenza vaccination observed for both immunocompetent AND immunocompromised patients.  Given that influenza vaccination, even when it fails to prevent laboratory-confirmed influenza, has still been associated with reductions in hospitalization, ICU admission, and death in several studies, we really need to be more on top of vaccinations for higher-risk immunocompromised patients. 31298691

In trying to reduce intestinal ESBL carriage in cancer patients, we accidentally gave them colistin-resistant organisms. Whoops!  There are now at least three approaches to decolonizing patients with specific pathogens at specific sites – antiseptics (e.g. chlorhexidine baths and mouthwash), antibiotics (e.g. nasal mupirocin for MRSA), and microbiota transplantation (e.g. FMT for intestinal carriage of MDR bacteria).  The FDA recently released a cautionary advisory and asked FMT investigators to review their screening practices following the death of one patient and development of a severe infection by another, both due to ESBL E.coli after FMT.  This paper reminds us that antibiotic decolonization poses pitfalls as well.

The CLEAR trial randomized patients with intestinal colonization with ESBL E.coli and K.pneumo (ESBL-EC/KP) 2:1 to receive either a 7-day regimen of oral colistin, gentamicin, and fosfomycin, or placebo.  The researchers collected fecal, urine, and throat samples on enrollment and at weeks 1, 2, 4, and 6 after treatment, then quantified the burden of ESBL-EC/KP, presence of resistance genes, and shifts in the intestinal microbiota after treatment.  The primary endpoint was eradication of ESBL-EC/KP from the intestine. 

The study was terminated prematurely after colistin powder’s production was suspended worldwide, and this paper reports the results of the 29 patients who were enrolled (n=18 treatment, n=11 placebo).  Patients’ baseline characteristics, including rates of neutropenia and exposure to chemotherapy and immunosuppressives, were similar between the groups.  Four patients in the treatment group stopped the study early due GI side effects, and one patient in the placebo group stopped due to increase in creatinine.  At weeks 1-2, more patients in the treatment group were no longer colonized (61% vs 18%; RR 3.36 with 95% CI 0.9-12.4), but this effect disappeared by weeks 4-6 (39% vs 27%; RR 1.4 with 95% CI 0.5-4.4).  However, fluoroquinolone resistance genes were more often detected in the post-treatment stool samples from patients who received antibiotics versus placebo, and two patients in the treatment group had post-treatment stools harboring the colistin resistance gene MCR-1.

So: oral antibiotics for intestinal decolonization produced only a transient reduction in ESBL-EC/KP carriage, and came at the expense of colonization with colistin-resistant organisms.  While it didn’t seem that the MCR-1 genes from those patients were present in the ESBL-EC/KP strains they carried, it’s reasonable to assume that adding this sort of selective pressure would let such an organism easily proliferate throughout cancer centers and wards.  No thanks! 31220256