Key message

Molecular diagnostics can improve diagnostic sensitivity of febrile episodes in children with cancer and help rationalize the management.

Introduction

Most cases of fever in children with cancer are treated empirically with antibiotics1 and nearly three-fifths of febrile episodes without a focus or documented infection have a respiratory pathogen identified in nasopharyngeal (NP) swab. For this reason, the present study was performed which focuses on molecular identification of the respiratory pathogens in samples collected from pediatric cancer patients presenting with fever with or without respiratory illness.

Methodology

This is a prospective observational study to identify pathogenic etiology of febrile episodes in pediatric cancer patients when on chemotherapy and study their clinical outcomes.

Results

There were 40 episodes of febrile illness during the 15 months among 27 patients on therapy for cancer. There was no gender predilection (male:female 14:13). The age group ranged from 1 year 6 months to 11 years 4 months (mean and median age group of 5 years). Majority of the patients had hematological malignancy (95%) as their primary diagnosis. Half of the febrile episodes were observed in children less than 5 years.

NP swab was positive in 28 out of 40 episodes (70%) but in 12 episodes (30%) no pathogen was identified. Seasonal predilection to monsoon and winter season for admission in the hospital was evident in those with positive NP swab. There were 26 out of 28 episodes admitted from September to March (92.85%) (inclusive of both years—2019 and 2020; Fig. 1).

Close

Fig. 1: Seasonal distribution of respiratory pathogens.

Export to PPT

Respiratory symptoms were present in 19 out of 28 episodes with positive NP swab (67.86%), but only in 6 out of 12 episodes with negative NP swab (50%; Table 1). The duration of hospitalization was comparatively lesser in NP-swab–positive isolates than in negative swabs though not statistically significant. More than 50% (16/28) had absolute neutrophil count (ANC) <1,000 and there were six episodes with ANC <100 in positive the NP swab category (21.42%). Whereas only 2/12 (16.66%) and 5/12 (41.7%) in the negative NP swab category had ANC <100 and ANC <1,000, respectively. The duration of antibiotics was more or less the same in both NP-swab–positive and negative episodes (range: 7.53 ± 2.79 in positive and 8.75 ± 3.57 days in negative episodes). None of the episodes were severe enough for intensive care unit (ICU) admission or oxygen requirement. One episode of negative NP swab required ICU admission 10 days after hospitalization in view of persisting fever and manifestation of central nervous system symptoms. All the above categories of comparison like the respiratory symptoms, duration of hospitalization, duration of antibiotics, and need for ICU admissions or oxygen requirements were statistically not significant.

Among episodes with positive NP swab, the organisms identified were single respiratory virus in 15/28 (53.57%), multiple viruses in 6/28 (21.42%), respiratory bacteria in 4/28 episodes (14.28%), and co-infection of both respiratory virus and bacteria in 3/28 (10.71%). Majority of viral episodes were symptomatic (18/28, 64.29%), irrespective of whether the NP swab showed a single virus (12/15, 80%), multiple viruses (4/6, 66.67%), or viral co-infection with bacteria (2/3, 66.67%). In contrast, isolated bacterial NP swabs were largely without respiratory symptoms (1/4, i.e., only 25% with symptoms).

Out of the single organisms isolated through NP swab (virus: 15 and bacteria: 4), the most common was Rhinovirus (RV; n = 9/19; 47.36%). Next in the prevalence was Streptococcus pneumoniae (n = 4/19; 21.05%) followed by human parainfluenza virus 3 (HPIV3; n = 3/19; 15.78%). Among the multiple organisms isolated, there were six episodes with multiple viruses and three with combination of virus and bacteria (Table 2).

On comparing the three most prevalent organisms identified in our study, we found that RV and HPIV3 were prevalent in wider age groups starting from 1 year 5 months to 8 years, whereas S. pneumoniae was prevalent among 4-to-6-year age groups (Table 3). Of all the variables, respiratory symptoms seem to be significant for viruses (RV and HPIV3) than bacterial isolate (S. pneumoniae). None of the three organisms required prolonged hospitalization or oxygen therapy or ICU admission. The most commonly used antibiotic as an empiric management as per febrile neutropenia institutional protocol was piperacillin–tazobactam and aminoglycoside was added when there was a clinical suspicion of resistant organism.

Discussion

Fever in children with cancer during chemotherapy is quite common owing to their immunocompromised status.23 This is one of the main causes of morbidity, delay in chemotherapy schedule, and sometimes mortality.4 These patients often are initiated on broad-spectrum antibiotics to cover serious bacterial diseases.5 Decision on discontinuation of antibiotics is also subjective6 and hence adds up to indiscriminate use of antibiotics1 and potential emergence of antibiotic resistance.5

The respiratory tract of immunocompromised patients becomes a vulnerable target for microorganisms, pathogenic or opportunistic, as it is directly connected with the environment.6 In healthy children, respiratory viruses are usually confined to the upper respiratory tract; in immunocompromised patients, owing to defects in innate and adaptive immunity and frequent visits to hospital, respiratory infections can rapidly disseminate6 and progress to the lower respiratory tract leading to respiratory distress and pulmonary compromise.5

Routine chest X-ray is not recommended in the initial assessment of pediatric febrile neutropenia2 in the absence of respiratory symptoms. In recent years, significant progress has been implemented in molecular techniques for the detection of respiratory pathogens as compared with conventional methods of cell culture or antigen detection.5 Although there are several studies on respiratory organisms in children with cancer, there are limited reported studies on their detection by multiplex molecular amplification assays (RT-PCR) in India.17 Our study analyzed the RT-PCR positivity and correlated it with the respiratory illness of patients with febrile episodes with or without neutropenia.

Our study could identify pathogens (either single virus or bacteria or multiple viruses or co-infection) among 70% subjects through this technique, whereas it was 46% in Torres et al8 and 52% in Suryadevara et al's9 studies. The reason for higher percentage in our study could be because we tested for a larger panel of 33 respiratory pathogens unlike other studies. In our study, half of the febrile episodes were observed in children less than 5 years, which is almost similar to the study done by Soudani et al.5

RV was the most commonly identified pathogen in our study, similar to few other studies.18910 The next common organism was S. pneumoniae in contrast to other studies that had RSV as the second most common virus isolate.810 The reason could be that these studies had only viruses in the respiratory panel unlike our study which included bacteria also in the respiratory panel. We could also find that there was co-presence of respiratory organism (either single virus or a co-infection) detected in more than half of the febrile episodes with culture-positive bacteremia. Since we needed to exclude the patients who had a microbiologically documented infection or focus, we did not take them into our analysis.

We found that respiratory symptoms were present at admission in 67.86% of the episodes with NP swab positive for respiratory virus or bacteria, which corresponds to studies done by Torres et al8 and Suryadevara et al.9 We also noticed that those with positive NP swab for RV had obvious respiratory symptoms compared to those with positive NP swab for S. pneumoniae or HPIV3, which was a statistically significant finding in our study. Similar to the study done by Srinivasan et al,10 we had more cases from September to March.

In our study, the duration of hospitalization was comparatively low in NP-swab–positive episodes than in NP-swab–negative episodes, though it was statistically not significant. None of the episodes had any adverse outcomes or ICU admissions. Likewise, in the study by Torres et al,3 all children with a positive respiratory virus detection, in the absence of a positive bacterial detection, had a favorable outcome.

The increasing detection of respiratory organisms highlights the importance of optimizing molecular diagnosis as part of the workup in cancer children when admitted for febrile episodes while on chemotherapy. It, however, needs to be analyzed if the PCR positivity actually represents the direct causative agent of the neutropenic fever or not, because a positive PCR result could also represent a subclinical infection, a postinfection viral shedding, or just intracellular nonreplicating viral nucleic acid remnants. Limitations of this study were we had a small sample size and since all children admitted were invariably initiated on empirical antibiotics as per febrile neutropenia institutional protocol, we could not evaluate the clinical course of positive NP swabs in the absence of antibiotic use. Furthermore, we also did not involve any control group to conclude if the NP swab results are to be taken as an acute infection or as commensals in the throat.

Conclusion

In our study, the PCR-based diagnostics could identify around 70% of respiratory pathogen in febrile episodes without a focus or documented infection. PCR-based diagnostics for respiratory pathogens can greatly enhance the management of patients with febrile neutropenia where standard tests show no documentation of cause of the infection. In the future, application of agile and broad microbial diagnostics may provide decreased use of antibiotics and individualize infection management.