Procalcitonin has been studied as a marker of infection to in patients of febrile neutropenia. To the best of our knowledge, we are aware of only 4 studies who have studied procalcitonin serially (Table 4). Each of those studies had a different design and hence their outcomes too have been different. Also, the cut offs for PCT in each of these studies have been different.
Most of our patients were high risk subjects of febrile neutropenia which is supported by the fact that 77% of our febrile episodes included patients of acute leukemia. We purposefully excluded patients who had fever during ATG therapy and Hematopoietic stem cell transplant (HSCT) patients as we were aware of reports ATG leading to falsely high PCT values. Hatzistilianou et al study included only children with acute lymphoblastic leukemia. Whereas, Fleishhack et al and Reitmann et al have included not only neutropenic children but also non-neutropenic children in their study to compare the levels between the two groups. Robinson et al have included patients with autologous HSCT subjects as well in their study [9]. The median age of the cohort in our study was 46.5 years which is much less when compared to the cohort in Robinson et al where the median age of the subjects was 57 years [9]. One plausible explanation for this observation could be because of 46% febrile episodes in their study included subjects of myeloma/lymphoma which is usually the disease of the elderly.
We had a significant burden of gram-negative infections with most common being Klebsiella pneumonia (37.5%) and significant number of patients fell into 2 or more groups of infections. E. coli (27.5%) followed by other gram-negative bacteria (Acinetobacter spp., Pseudomonas) were the most common organism. [17-18] followed by gram positive organisms. This is comparable to the data that was published. But in the study by Rasmy et al, it is quiet contrasting where they found gram positive organisms to be the predominant bacteria responsible for febrile neutropenia [19].
The median duration of neutropenia in our study was 36 days (range=2-400) and the median duration of hospitalization prior to febrile event in our study was 15(range=2-90) days which contrasts with the study by Robinson et al. whose median duration of neutropenia was 15 days and duration of hospitalization was about 28 days [9]. Probable reason for this observation could be referral bias among patients of Aplastic anemia and acute leukemia.
27 % of the febrile episodes were the 2nd febrile episode whereas by in the Robinson et al study, 31% of their febrile events were 2nd febrile episodes. The number of first febrile event and 2nd febrile events are comparable between both the studies. Similar data with respect to febrile episodes are not available in the other two studies.
Hypoxemia requiring mechanical ventilation/ septic shock was seen in 39(69%) of all the infections. Which contrasts with the study by Robinson et al. where only 2% of their patients were critically ill. This is most likely because most of our infections were caused due to gram-negative bacteria and they are known to cause severe sepsis.
The median highest PCT values were seen 48 hours after onset of fever in the CDI group. When the CDI was having a MDI(n=9) too, the median PCT values were highest on day 14. It means the CDI group responded to the empirical antibiotic regimen and hence within 48 hours the PCT values started to fall. But when CDI was associated with an MDI the antibiotic regimen was not appropriate and hence the PCT values kept rising and had a late response when the antibiotics were changed. Reitman et al compared PCT values which was restricted to studying only bacteremia, they did not include any group with CDI [16].
PCT values in the MDI group peaked on day 7. PCT seems to be an exquisitely specific marker of bacteremia because irrespective of presence of any other cause of fever PCT values were highest on day 7 in MDI when compared to the PUO/CDI (only)/PUO groups. Robinson et al in their study found that the PCT values were significantly higher in non-CNS MDI when compared to all other etiologies of fever [9]. They found that the peak PCT was seen 2 days after onset of fever which was usually the time, they re-evaluated their empirical antibacterial therapy. At our center our antibiotic policy is to change the antibiotics at 48- 72 hours of fever not responding to empirical antibiotics and the response was seen about 48 to 72 hours after reviewing the first line antimicrobials.
The PCT values in our study rose to their peak at 48 hrs in the IFI group. The peak PCT in the IFI group was not statistically different from the MDI group. Most of the IFI s in our study were of probable fungal infection, proven fungal infections were seen in only 3 cases (all were Aspergillus spp.) and no case of invasive candidiasis was found. A PCT value at 24 hours of ≤1.2 with AUC of 0.766 had a sensitivity and specificity of 62.5% and 87.5% for discriminating IFI and MDI (p=0.033). Also, PCT had a negative predictive value of 70% for the diagnosis of IFI when compared to MDI.
When PCT was compared to PUO at day 7, a value of 0.4 with AUC of 0.813 had a sensitivity and specificity of 60.98% and 100% for discriminating IFI and PUO with statistically significant value of 0.0001.
Robinson et al in their study found that a PCT >500pg/ml beyond 3 days of fever had a sensitivity of 81%, specificity of 57%, PPV 49%, NPV 86% to diagnose fungal infection and it occurred around day 7 in their study [9].
The MDI+IFI is a unique group which has not been highlighted in other studies. Though the numbers are small (n=3) but this has not been forthcoming in any other studies.
At fever onset the median PCT was highest in this group but it was not statistically significant. There was no statistically significant difference in the PCT values at other time points when compared to other infections.
There was a significant fall in the level of PCT by day 14 in the group of patients who had both IFI+ MDI when compared to other groups (p=0.047) which means that we were able to successfully identify and treat these group of patients. Fleishhack et al. too found in their study that decreasing PCT levels during febrile events reflected defervescence, clinical improvement [14]. Robinson et al. in their study found that a decreasing PCT to normal levels was significantly associated with defervescence and response to antifungal therapy when there was persistently elevated PCT values [9].
The median PCT values were significantly higher in the group undergoing salvage chemotherapy as it is expected that these patients have come with either a bacterial/fungal infection prior to starting chemotherapy and have dysfunctional granulocytes which predispose them to infection. Robinson et al. in their study did not find any significant difference in the PCT values when subjects were stratified to the type of chemotherapy they received [9].
Procalcitonin has been studied as a marker of infection to in patients of febrile neutropenia. To the best of our knowledge, we are aware of only 4 studies who have studied procalcitonin serially (Table 4). Each of those studies had a different design and hence their outcomes too have been different. Also, the cut offs for PCT in each of these studies have been different.
Most of our patients were high risk subjects of febrile neutropenia which is supported by the fact that 77% of our febrile episodes included patients of acute leukemia. We purposefully excluded patients who had fever during ATG therapy and Hematopoietic stem cell transplant (HSCT) patients as we were aware of reports ATG leading to falsely high PCT values. Hatzistilianou et al study included only children with acute lymphoblastic leukemia. Whereas, Fleishhack et al and Reitmann et al have included not only neutropenic children but also non-neutropenic children in their study to compare the levels between the two groups. Robinson et al have included patients with autologous HSCT subjects as well in their study [9]. The median age of the cohort in our study was 46.5 years which is much less when compared to the cohort in Robinson et al where the median age of the subjects was 57 years [9]. One plausible explanation for this observation could be because of 46% febrile episodes in their study included subjects of myeloma/lymphoma which is usually the disease of the elderly.
We had a significant burden of gram-negative infections with most common being Klebsiella pneumonia (37.5%) and significant number of patients fell into 2 or more groups of infections. E. coli (27.5%) followed by other gram-negative bacteria (Acinetobacter spp., Pseudomonas) were the most common organism. [17-18] followed by gram positive organisms. This is comparable to the data that was published. But in the study by Rasmy et al, it is quiet contrasting where they found gram positive organisms to be the predominant bacteria responsible for febrile neutropenia [19].
The median duration of neutropenia in our study was 36 days (range=2-400) and the median duration of hospitalization prior to febrile event in our study was 15(range=2-90) days which contrasts with the study by Robinson et al. whose median duration of neutropenia was 15 days and duration of hospitalization was about 28 days [9]. Probable reason for this observation could be referral bias among patients of Aplastic anemia and acute leukemia.
27 % of the febrile episodes were the 2nd febrile episode whereas by in the Robinson et al study, 31% of their febrile events were 2nd febrile episodes. The number of first febrile event and 2nd febrile events are comparable between both the studies. Similar data with respect to febrile episodes are not available in the other two studies.
Hypoxemia requiring mechanical ventilation/ septic shock was seen in 39(69%) of all the infections. Which contrasts with the study by Robinson et al. where only 2% of their patients were critically ill. This is most likely because most of our infections were caused due to gram-negative bacteria and they are known to cause severe sepsis.
The median highest PCT values were seen 48 hours after onset of fever in the CDI group. When the CDI was having a MDI(n=9) too, the median PCT values were highest on day 14. It means the CDI group responded to the empirical antibiotic regimen and hence within 48 hours the PCT values started to fall. But when CDI was associated with an MDI the antibiotic regimen was not appropriate and hence the PCT values kept rising and had a late response when the antibiotics were changed. Reitman et al compared PCT values which was restricted to studying only bacteremia, they did not include any group with CDI [16].
PCT values in the MDI group peaked on day 7. PCT seems to be an exquisitely specific marker of bacteremia because irrespective of presence of any other cause of fever PCT values were highest on day 7 in MDI when compared to the PUO/CDI (only)/PUO groups. Robinson et al in their study found that the PCT values were significantly higher in non-CNS MDI when compared to all other etiologies of fever [9]. They found that the peak PCT was seen 2 days after onset of fever which was usually the time, they re-evaluated their empirical antibacterial therapy. At our center our antibiotic policy is to change the antibiotics at 48- 72 hours of fever not responding to empirical antibiotics and the response was seen about 48 to 72 hours after reviewing the first line antimicrobials.
The PCT values in our study rose to their peak at 48 hrs in the IFI group. The peak PCT in the IFI group was not statistically different from the MDI group. Most of the IFI s in our study were of probable fungal infection, proven fungal infections were seen in only 3 cases (all were Aspergillus spp.) and no case of invasive candidiasis was found. A PCT value at 24 hours of ≤1.2 with AUC of 0.766 had a sensitivity and specificity of 62.5% and 87.5% for discriminating IFI and MDI (p=0.033). Also, PCT had a negative predictive value of 70% for the diagnosis of IFI when compared to MDI.
When PCT was compared to PUO at day 7, a value of 0.4 with AUC of 0.813 had a sensitivity and specificity of 60.98% and 100% for discriminating IFI and PUO with statistically significant value of 0.0001.
Robinson et al in their study found that a PCT >500pg/ml beyond 3 days of fever had a sensitivity of 81%, specificity of 57%, PPV 49%, NPV 86% to diagnose fungal infection and it occurred around day 7 in their study [9].
The MDI+IFI is a unique group which has not been highlighted in other studies. Though the numbers are small (n=3) but this has not been forthcoming in any other studies.
At fever onset the median PCT was highest in this group but it was not statistically significant. There was no statistically significant difference in the PCT values at other time points when compared to other infections.
There was a significant fall in the level of PCT by day 14 in the group of patients who had both IFI+ MDI when compared to other groups (p=0.047) which means that we were able to successfully identify and treat these group of patients. Fleishhack et al. too found in their study that decreasing PCT levels during febrile events reflected defervescence, clinical improvement [14]. Robinson et al. in their study found that a decreasing PCT to normal levels was significantly associated with defervescence and response to antifungal therapy when there was persistently elevated PCT values [9].
The median PCT values were significantly higher in the group undergoing salvage chemotherapy as it is expected that these patients have come with either a bacterial/fungal infection prior to starting chemotherapy and have dysfunctional granulocytes which predispose them to infection. Robinson et al. in their study did not find any significant difference in the PCT values when subjects were stratified to the type of chemotherapy they received [9].