Baseline Inflammation Reordered Pneumonitis Risk in Esophageal Cancer
In esophageal squamous cell carcinoma, the combination of immune checkpoint inhibitors and thoracic radiotherapy creates a familiar oncologic tradeoff: stronger local and systemic control, but a lung toxicity profile that can complicate treatment continuity. A single-center retrospective cohort from Hebei University examined whether baseline systemic inflammation changed the relationship between lung radiation dose and subsequent checkpoint inhibitor pneumonitis in 268 patients treated with PD-1 or PD-L1 blockade. The answer was not simply that radiation increased risk. The signal was more selective than that, and it sharpened when systemic inflammation was already high.
Grade ≥2 pneumonitis was the prespecified primary endpoint, and it occurred in 29 of 268 patients overall. Grade ≥1 events reached 77 patients, while grade ≥3 events occurred in 19. In the radiotherapy subgroup, pneumonitis was more common still: 54 of 154 patients developed any-grade disease, 20 developed grade ≥2 disease, and 13 developed grade ≥3 disease. Every pneumonitis event in the radiotherapy subgroup began after radiotherapy had ended, which argues against simple overlap with acute radiation injury and supports the clinical challenge of separating delayed immune toxicity from dose-related pulmonary vulnerability.
High SII Marked a More Fragile Pulmonary Risk Profile
Baseline systemic inflammation was captured with the systemic immune-inflammation index, calculated from a complete blood count drawn within 7 days before the first immunotherapy dose. The optimal cutoff for clinically significant pneumonitis was 684.22, identified using the Youden index. In the full cohort, this threshold separated 170 patients into a low-SII group and 98 into a high-SII group. The AUC was 0.72, with sensitivity of 76.5% and specificity of 68.2%.
Higher baseline SII tracked with more severe pneumonitis grades. That pattern held in the overall cohort and in the radiotherapy subgroup, where higher severity thresholds were associated with higher SII distributions. The biologic plausibility is straightforward: elevated neutrophil and platelet burden alongside relative lymphopenia points to a pro-inflammatory, immunologically stressed host state. In that setting, the lung may have less reserve against a second hit from thoracic radiation, especially when checkpoint blockade has already shifted immune activation toward a more reactive state.
COPD also entered the model as an independent risk factor, which is clinically intuitive given baseline airway injury and reduced pulmonary reserve. Thoracic radiotherapy itself independently increased risk, while higher BMI appeared protective. In bedside terms, the model sketches a patient who is not just receiving treatment, but receiving it on top of a lung that may already be inflamed, structurally compromised, or less able to tolerate epithelial injury.
The Dose Relationship Tightened in the High-SII Group
The most clinically interesting finding came from the interaction analysis between baseline SII and lung V20 in the radiotherapy subgroup. V20, modeled per 5-percentage-point increase, was associated with CIP, but the association was concentrated in the high-SII stratum. Formal SII-by-V20 interaction testing was significant across grade ≥1, grade ≥2, and grade ≥3 endpoints.
This does not establish a new universal V20 limit. It does suggest that the same dosimetric exposure may not carry the same biological meaning in every patient. In a high-inflammatory host, the lung’s response to radiation may be amplified by immune priming, endothelial activation, and impaired inflammatory resolution. The dose-volume histogram, in other words, may need to be read alongside the patient’s baseline inflammatory state rather than in isolation.
Among radiotherapy patients, the median V20 was 18.60%, the median mean lung dose was 6.11 Gy, and the median follow-up after radiotherapy initiation was 16.2 months. The exploratory Kaplan–Meier cutoff of 17.66% was retained for visualization only and was not treated as a validated clinical threshold. That caution is appropriate. A clean threshold in a single cohort can look more decisive than the underlying biology actually is.
The Nomogram Performed Acceptably, but It Remains Internal
A grade ≥2 pneumonitis nomogram incorporated BMI, COPD, thoracic radiotherapy status, and baseline SII category. Internal validation suggested the model was reasonably stable: the C-index was 0.758, the mean 10-fold cross-validated C-index was 0.74, and bootstrap optimism correction yielded 0.73. Calibration and decision-curve analyses supported acceptable internal performance and potential clinical utility.
The practical appeal of the nomogram is clear. It moves beyond a single-variable mindset and combines host factors with treatment exposure. A patient with COPD, high SII, and thoracic radiotherapy is not interchangeable with a patient who lacks those features, even if their cancer stage and systemic therapy backbone look similar on paper. The model is still hypothesis-generating, but it gives clinicians a more realistic way to think about pneumonitis risk than dose alone.
What the Cohort Can and Cannot Support
This was a retrospective single-center study, and that design matters when translating a toxicity signal into practice. The cohort was limited to 268 patients, all from one hospital, and external validation was not available. The investigators also noted that only 14 radiotherapy patients received PD-L1 inhibitors, making separate estimates for that subgroup statistically unstable. Patients received immune checkpoint inhibitors, and 154 later underwent thoracic radiotherapy after 2 to 4 induction cycles; immune therapy was withheld during radiotherapy and not resumed afterward.
Those details matter because they define the treatment context in which the findings apply. The study does not justify a new lung constraint, and it does not prove causality. What it does show is a reproducible-looking interaction between baseline systemic inflammation and radiation dose exposure, with worse pneumonitis risk concentrated in the high-SII subgroup. For esophageal cancer teams balancing immune therapy and thoracic radiotherapy, the main clinical message is narrower: inflammatory status may identify the patients in whom the lung dose curve becomes steeper, even before symptoms start.
The next step is external validation in independent cohorts, followed by mechanistic work that can explain why systemic inflammation and lung dosimetry converge on pneumonitis risk so unevenly across patients.