Infectious complications remain a major cause of morbidity and mortality following aggressive chemotherapy or hematopoietic cell transplantation. Early appropriate treatment decreases mortality. C-reactive protein (CRP) is a biomarker of infections, however data whether CRP can help in early detection of infections in patients undergoing hematopoietic cell transplantation are scarce.
We aimed to prospectively examine whether CRP elevation precedes the clinical signs and symptoms of sepsis in HCT recipients with bloodstream infections (BSIs). In all patients in the hematopoietic cell transplantation ward, CRP blood levels were measured once daily. Cases were defined as patients with clinically-significant BSI (bacteremia or fungemia). Controls were defined as afebrile patients without a suspected or document infection. Cohorts were matched by the type of transplantation, time from transplantation, GVHD status and age. We calculated the mean difference (MD) between the CRP on day -1 (1 day before clinical suspicion of infection) and days -2 and -3 (deltaM1M2 and delta M1M3, respectively). We compared these delta values between cases and controls. Using the delta M1M value observed in our case-control study, we examined the sensitivity, specificity, positive and negative predictive values of this difference in CRP between consecutive days for diagnosis of BSIs among all admitted patients.
From January 2010 to April 2012 we identified 50 cases of BSI (gram negative, n=32; gram positive, n=15 and yeasts, n=3). Median days from HCT to BSI onset was 78 (range, (−2) – 232). 46 episodes (92%) occurred in patients receiving allografts and among these patients, in 39 patients (85%), episodes were preceded by graft vs. host disease. Delta M1M3 and delta M1M2 were significantly higher in patients with BSI compared to controls (MD=4.47, 95% CI 2.2–6.7, p<.001 and MD=3.7, 95% CI 1.99–5.4, p<.001, respectively). Only a trend for increased delta CRP was observed between days -2 and -3 (MD=1.4, 95% CI 0.9–2.9, p=.07). In the overall cohort, sensitivity and specificity of a daily delta value >4 were 52% and 98%, respectively. The positive and negative predictive values were 66% and 98%, respectively. Other etiologies that were associated with a daily increase in CRP >4 were non-bacteremic infections, infusion reactions associated with administration of anti-thymocytes globulin, engraftment and the onset of graft vs. host disease.
We conclude that CRP has a valuable role in the sequential monitoring of patients undergoing hematopoietic cell transplantation and might serve as an early predictor for infectious episodes. The high negative predictive value of CRP in decision analysis regarding empiric antibiotic treatment in cases of fever should be further studied. Increase CRP blood levels in afebrile HCT recipients should raise the suspicion for infectious etiology.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.