Postoperative nutrition decline may flag infection after oral cancer surgery

A rapid decline in postoperative nutritional status may help identify patients at elevated risk of surgical-site infection after surgery for oral cancer, according to a retrospective study from researchers at the Saitama Cancer Center in Japan. The investigators found that a fall in the prognostic nutritional index between the first and third days after surgery was associated with infection at the reconstructed recipient site. Their analysis also revealed an important limitation: the association was statistically significant, but the marker was only moderately able to distinguish infected from uninfected patients and had not yet been tested in an independent population.

The study included 171 patients who underwent oral cancer resection with reconstruction using either a free flap or a pedicled flap between September 2012 and April 2026. Surgical-site infection developed in 33 patients, corresponding to 19.3% of the cohort, and the earliest case was diagnosed on postoperative day 6. The researchers focused on the prognostic nutritional index, or PNI, a measure of nutritional and immune status that is calculated from blood biomarkers. They assessed PNI on postoperative days 1, 3 and 5, then examined two within-patient changes: the ratio on day 3 divided by the value on day 1, and the ratio on day 5 divided by the value on day 3.

Patients who later developed an infection had a lower median day-3-to-day-1 PNI ratio than those who remained infection-free: 0.87 compared with 0.94, a difference with a P value of 0.006. A ratio below 1 therefore represented a decline in PNI over the first three postoperative days, while a ratio above 1 indicated improvement or no decline. Because the study tracked changes within individuals rather than relying only on a single laboratory value, the result suggests that the direction of a patient’s early recovery may carry information that is not visible in one measurement alone.

The investigators used landmark-based Cox regression to estimate the association between these postoperative ratios and infection through postoperative day 30. In a model established at the day-5 landmark and adjusted for the day-5-to-day-3 PNI ratio, operative time and diabetes mellitus, each 0.1 increase in the day-3-to-day-1 ratio was associated with a hazard ratio of 0.635 for surgical-site infection. In statistical terms, that corresponds to a 36.5% lower instantaneous risk of infection for every 0.1-unit rise in the ratio, with a 95% confidence interval from 0.423 to 0.951 and a P value of 0.027. The confidence interval indicates uncertainty around the estimated effect, while the P value indicates that the association was unlikely to be explained by chance alone under the statistical model.

A more clinically relevant model used only information available by postoperative day 3. After adjustment, each 0.1 increase in the day-3-to-day-1 PNI ratio was associated with a hazard ratio of 0.585 for infection, equivalent to a 41.5% lower instantaneous risk. Its 95% confidence interval ranged from 0.398 to 0.860, and the P value was 0.006. This model is particularly notable because it could generate a risk estimate without waiting for postoperative day 5 data. The earliest infection in the cohort occurred on day 6, yet the day-3 landmark was set sufficiently early to precede the first observed event and allow the subsequent infection risk to be estimated.

Although the statistical association was clear, the marker’s ability to classify patients was described as modest. The adjusted day-5-landmark model had an apparent concordance index of 0.682, which fell to 0.646 after correction for optimism using bootstrap resampling. The day-3-to-day-1 PNI ratio itself produced an area under the receiver operating characteristic curve of 0.655. An area under the curve of this size is better than chance but not strong enough by itself to provide highly reliable individual predictions. Performance could also vary depending on the threshold selected and the clinical setting.

At a cutoff of 0.913 derived from the study cohort, the day-3-to-day-1 ratio had a positive predictive value of 30.1% and a negative predictive value of 90.9%. A positive result at this threshold therefore did not mean that an infection was highly likely: only about three in ten patients who crossed the threshold subsequently developed one. By contrast, the high negative predictive value meant that most patients without the low-ratio pattern remained free of infection during follow-up. These values depend on the prevalence and characteristics of this particular cohort, and the threshold has not been externally validated.

The researchers also compared the PNI ratio with postoperative markers already used in clinical assessment. Its area under the curve was not detectably different from that of the albumin day-3-to-day-1 ratio, which reached 0.664; the paired P value was 0.822. The comparison did not establish that either marker was superior, and it also did not show that the PNI offered independent value beyond albumin. The authors used sensitivity analyses and paired comparisons to assess the robustness and clinical context of the findings, but the results still point to a potential warning signal rather than a definitive diagnostic test.

Several features of the research limit how far the findings can be generalized. The analysis was retrospective and conducted at a single center, meaning that local practices, patient characteristics and infection-control procedures may have influenced the results. The sample was relatively modest for a prognostic study, with 33 infections providing the events used to estimate risk. Early postoperative changes could also reflect the physiological stress of major surgery, inflammation, fluid shifts, differences in postoperative nutrition or other clinical changes, so the PNI trajectory may be a marker of surgical stress as well as a direct measure of infection risk. The retrospective design cannot establish causation.

For clinicians, the immediate implication is that the first three days after oral cancer resection may contain useful information about a patient’s trajectory. A sharp decline in the prognostic nutritional index could identify a group needing closer surveillance, continued nutritional support and attentive wound assessment. Yet the study does not justify using a single cutoff to start antibiotics, change flap management or make other treatment decisions. Infection assessment still requires clinical evaluation, examination of the wound and appropriate microbiological testing. Before the PNI ratio or its cohort-derived threshold can enter routine decision-making, it needs prospective evaluation and external validation in larger and more diverse populations, followed by testing of whether flagging risk actually improves outcomes.

Subject of Research: Association of early postoperative prognostic nutritional index changes with surgical-site infection after oral cancer resection and flap reconstruction

Article Title: Association of early postoperative changes in the prognostic nutritional index with surgical site infection after oral cancer resection with flap reconstruction

Article References: Masuda, H., Sumino, J., Katsurano, M., Yoshida, K., Shimojukkoku, Y., Hamahata, A., Konoeda, H., Tazaki, A., Futami, H., Tanaka, S., & Yagihara, K. (2026). Association of early postoperative changes in the prognostic nutritional index with surgical site infection after oral cancer resection with flap reconstruction. BMC Cancer. https://doi.org/10.1186/s12885-026-17212-7

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17212-7

Keywords: oral cancer, surgical-site infection, prognostic nutritional index, flap reconstruction, postoperative monitoring, nutrition, immunonutritional status, wound infection, clinical risk prediction, Association, early, postoperative