One in Five Lymphoma Patients on R-CHOP Suffers Febrile Neutropenia, Real-World Analysis Finds

Every year, hundreds of thousands of people worldwide are diagnosed with diffuse large B-cell lymphoma, the most common form of non-Hodgkin lymphoma, and nearly all of them begin treatment with the same backbone: R-CHOP, a combination of the antibody rituximab and four chemotherapy drugs that has anchored first-line therapy for more than two decades. The regimen is effective, but it carries a well-known hazard. By wiping out neutrophils, the white blood cells that form the first line of defense against bacteria, chemotherapy opens the door to febrile neutropenia, a fever during a state of profound immune suppression that is treated as a medical emergency because it can escalate to sepsis and death within hours.

For years, clinical guidelines have drawn a bright line at twenty percent. If a chemotherapy regimen is expected to cause febrile neutropenia in at least one in five patients, guidelines from the American Society of Clinical Oncology, the European Organisation for Research and Treatment of Cancer, and the National Comprehensive Cancer Network recommend giving granulocyte colony-stimulating factor, or G-CSF, prophylactically from the first cycle. This injected growth factor stimulates the bone marrow to keep producing neutrophils through the nadir of each chemo cycle, and randomized evidence shows it reduces both febrile neutropenia and chemotherapy-associated mortality. The trouble is that the twenty percent figure for R-CHOP has rested largely on trial data, and a new systematic review argues that those data may systematically understate what actually happens in ordinary clinics.

The review, published in Supportive Care in Cancer by Jamiel Reyes and Alexander Reyes, took an unusually careful approach to this question. The authors searched MEDLINE, CENTRAL, ClinicalTrials.gov, Europe PMC, Epistemonikos, Lens.org, Google Scholar, and citation trails from inception through 8 September 2026, looking for studies that reported febrile neutropenia incidence in adults with diffuse large B-cell lymphoma receiving first-line R-CHOP or R-CHOP-like immunochemotherapy. The protocol was prospectively registered in PROSPERO, and the team assessed risk of bias with the ROBINS-I tool for observational studies and RoB 2 for randomized trials, then graded the certainty of the evidence using the GRADE framework.

Fifteen studies covering 5,440 patients and 885 febrile neutropenia events made the final analysis. But the authors did something that many meta-analyses of this kind skip: they stratified everything by study design, pooling observational cohorts separately from clinical trials. The rationale is technical but important. Trials and routine-practice cohorts measure the same complication under materially different conditions, and blending them can produce a single number that describes neither setting well. In trials, patients tend to be fitter, monitoring is more intensive, supportive care protocols are more standardized, and prophylactic G-CSF use is often higher and more consistent than in the community.

The observational stratum told a strikingly consistent story. Across ten cohorts enrolling 4,020 patients, the pooled incidence of febrile neutropenia was 19.0 percent, with a 95 percent confidence interval of 17.1 to 21.1 percent. Heterogeneity was low, with an I-squared statistic of just 34 percent, meaning the studies agreed with one another far more than is typical in meta-analyses of this kind. The prediction interval, which estimates where a future study would likely fall, ran from 15.3 to 23.3 percent, and the estimate proved stable across every sensitivity analysis and leave-one-out test the authors performed, including a prespecified analysis restricted to the nine cohorts using core R-CHOP. In other words, no single study was driving the result.

The trial stratum looked very different. Five clinical trials enrolling 1,420 patients yielded a pooled febrile neutropenia incidence of 11.7 percent, but with a confidence interval stretching from 5.7 to 22.7 percent and an I-squared of 79.4 percent, indicating severe heterogeneity. Individual trial estimates ranged from 2.4 to 20.0 percent, a nearly tenfold spread that makes the pooled figure difficult to interpret. The formal comparison between the two designs did not reach conventional statistical significance, with a p-value of 0.051, but the authors note that the difference depended substantially on a single trial. When the large 2013 trial by Cunningham and colleagues comparing 14-day and 21-day R-CHOP schedules was omitted, the trial-stratum estimate shifted to 15.5 percent and the gap between designs essentially disappeared.

What does this mean for the twenty percent threshold? The observational estimate of 19.0 percent sits squarely at, not below, the conventional trigger for primary G-CSF prophylaxis, and its prediction interval crosses well above it. The authors’ conclusion is blunt: in routine practice, approximately one in five adults with diffuse large B-cell lymphoma receiving first-line R-CHOP-based therapy develops febrile neutropenia, and trial-derived estimates should not be assumed to represent this real-world burden. For clinicians who have been reassured by trial figures suggesting the risk is closer to one in ten, the message is that the community setting may be considerably riskier than the pages of a phase 3 protocol suggest.

The mechanistic reasons for the discrepancy are not hard to reconstruct. Trial populations are enriched for patients who meet strict eligibility criteria, often excluding older adults with multiple comorbidities, patients with poor performance status, and those with the very risk factors, such as advanced age, elevated lactate dehydrogenase, bulky disease, and low baseline blood counts, that observational studies have repeatedly linked to febrile neutropenia. Trials also tend to enforce prophylactic growth factor use more rigorously and to monitor neutrophil counts on tight schedules, catching problems before they become fevers. Real-world cohorts, by contrast, capture the full spectrum of patients, including elderly and frail individuals who make up a large share of diffuse large B-cell lymphoma diagnoses, and they reflect the uneven adoption of prophylaxis in everyday oncology.

The stakes extend beyond the acute emergency itself. Febrile neutropenia is one of the leading causes of chemotherapy dose reductions and treatment delays, and for a curable disease like diffuse large B-cell lymphoma, compromised dose intensity can translate directly into worse survival. Earlier research, including the INC-EU prospective observational study and analyses of R-CHOP delivery by Pettengell and colleagues, documented how neutropenic complications disrupt treatment schedules and drive hospitalizations. Prophylactic G-CSF, particularly the long-acting pegfilgrastim given once per cycle, has been shown to preserve relative dose intensity, and studies in patients aged 75 and older suggest it reduces mortality in this vulnerable group. If roughly one in five unselected patients experiences a febrile neutropenia event without adequate prophylaxis, the case for applying the guideline threshold generously, rather than treating it as a ceiling, becomes considerably stronger.

The review is not without caveats, and the authors are transparent about them. The between-design difference was not statistically significant at the conventional threshold, and the analysis was reported as an observation rather than a hypothesis test. The registered protocol underwent three amendments, including replacing a comparative outcome with patient-level febrile neutropenia incidence as the primary endpoint and adopting the design stratification during revision, changes the authors documented in the PROSPERO record. All fifteen studies were pooled irrespective of design in a secondary analysis, yielding 17.0 percent with high heterogeneity of 83.3 percent, a figure that illustrates precisely why the stratified approach matters: pooling across designs produces a number with a confidence interval so wide and an I-squared so high that it obscures the two very different realities it averages together. Still, the core observational finding is robust, low-heterogeneity, and consistent with the biological logic of the regimen. As newer first-line options such as polatuzumab vedotin-based therapy enter wider use and add their own myelosuppressive profiles to the mix, the review’s central lesson will only grow in relevance: when deciding who needs prophylactic growth factors, the numbers that count are the ones from the clinic down the street, not just the ones from the trial protocol.

Subject of Research: Febrile neutropenia incidence in adults with diffuse large B-cell lymphoma receiving first-line R-CHOP-based therapy

Article Title: Febrile neutropenia in adults with diffuse large B-cell lymphoma receiving first-line R-CHOP-based therapy: a systematic review and meta-analysis of observational and trial evidence

Article References: Reyes, J., & Reyes, A. (2026). Febrile neutropenia in adults with diffuse large B-cell lymphoma receiving first-line R-CHOP-based therapy: a systematic review and meta-analysis of observational and trial evidence. Supportive Care in Cancer, 34(10), Article 1065. https://doi.org/10.1007/s00520-026-11301-w

Image Credits: AI Generated

DOI: 10.1007/s00520-026-11301-w

Keywords: diffuse large B-cell lymphoma, febrile neutropenia, R-CHOP, G-CSF, primary prophylaxis, meta-analysis, systematic review, real-world evidence, supportive care, neutropenia, rituximab, clinical trials