Simple Blood Test Score Predicts Survival in Nonagenarians With Colorectal Cancer

Surgery for colorectal cancer in patients aged 90 and older has long been viewed with caution, if not outright skepticism. Advanced age was traditionally considered a relative contraindication to major abdominal operations, driven by fears of perioperative collapse in bodies with diminished physiological reserve. Yet surgical technique, anesthetic management, and postoperative care have improved to the point that nonagenarians are increasingly offered curative resection. The pressing question is no longer simply whether to operate, but how to identify which of these extraordinarily old patients will sail through surgery and which will struggle. A new multi-institutional study from Japan suggests that a remarkably simple calculation, built from a single blood draw and a height and weight measurement, may hold much of the answer.

The research, published in Annals of Gastroenterological Surgery, drew on the Setouchi Colorectal Neoplasm Registration database, which aggregates data from 15 hospitals affiliated with Okayama University. Between January 2011 and December 2022, the database identified 402 cases of colorectal cancer in patients aged 90 or above. After excluding patients lacking the nutritional measurements needed for analysis, 225 nonagenarians who underwent primary tumor resection remained in the cohort. The median age was 92 years, two-thirds were women, and nearly half carried an American Society of Anesthesiologists classification of 3, indicating severe systemic disease. This is, by any standard, a fragile population facing major surgery.

The tool at the center of the study is the Geriatric Nutritional Risk Index, or GNRI. It is calculated from serum albumin levels and body mass index using a straightforward formula: 14.89 times the albumin concentration in grams per deciliter, plus 41.7 times the BMI divided by 22. Albumin, the most abundant protein in blood plasma, reflects both nutritional intake and systemic inflammation, while BMI captures energy reserves. The index was originally developed to grade nutritional risk in hospitalized older adults and has since been validated as a prognostic marker across a range of surgical settings, including colorectal cancer resection in older patients. Its appeal lies in its simplicity: no specialized equipment, no subjective scoring, just two routinely collected values combined in seconds.

The researchers stratified their cohort using two cutoff values. The first, 98, comes from the existing literature and served as the prespecified threshold for the primary analysis. The second, 86, emerged from a post hoc receiver operating characteristic analysis of the cohort itself, identifying the value that best discriminated vital status at three years, with a sensitivity of 0.62 and a specificity of 0.55. By the literature-based cutoff, only 36 of the 225 patients qualified as nutritionally well; by the cohort-derived cutoff, the groups split almost evenly. The median GNRI across the whole cohort was 85.9, and median serum albumin was just 3.2 grams per deciliter, figures that underscore how prevalent malnutrition is in this age group even before cancer enters the picture.

On the short-term outcomes, the findings were nuanced. Postoperative complications of any severity, graded according to the Clavien-Dindo classification, occurred in 36.4 percent of patients, while severe complications of Grade III or higher affected only 5.8 percent. Thirty-day mortality was a strikingly low 0.9 percent, and only one patient required reoperation. After the investigators applied stabilized inverse-probability-of-treatment weighting, a propensity-score method that balances baseline characteristics such as age, sex, ASA class, pathological stage, and surgical approach between the GNRI groups, there was no significant difference in overall complication rates between nutritionally high and low patients at either cutoff. However, at the cutoff of 98, well-nourished patients did have a significantly shorter expected hospital stay, with a length-of-stay ratio of 0.70, meaning their stays were roughly 30 percent shorter than those of their nutritionally depleted counterparts.

The surgical patterns themselves revealed something important. Patients with high GNRI scores were significantly more likely to undergo laparoscopic rather than open surgery, had longer operative times, and required stoma creation far less often. The authors interpret this as clinical selection at work: frail, malnourished patients tend to have more comorbidities and more complex disease, prompting surgeons to favor open procedures perceived as safer, and to create protective stomas more liberally. This imbalance likely confounded short-term measures such as blood loss, which was significantly lower in the high-GNRI group at the 86 cutoff, and the researchers acknowledge that residual confounding cannot be excluded for these perioperative endpoints despite the weighting adjustment.

It is the long-term survival data that carry the most weight. In the weighted Cox proportional hazards analysis using the cohort-derived cutoff of 86, patients with high GNRI had a hazard ratio for death of 0.54 compared with low-GNRI patients, a 46 percent reduction in the risk of dying that was statistically significant. Median survival was 1,283 days in the high-GNRI group versus 700 days in the low-GNRI group, a difference of more than a year and a half. At the literature-based cutoff of 98, the same direction of effect was seen but did not reach significance in the full cohort, with a hazard ratio of 0.67. However, in a sensitivity analysis restricted to the 194 patients without metastatic disease, the adjusted hazard ratio at the 98 cutoff fell to 0.52 and became statistically significant, suggesting the survival advantage of good nutritional status is not merely an artifact of including patients with stage IV cancer.

Perhaps the most arresting detail lies in the causes of death. Of the 100 patients who died during follow-up, 18 deaths were definitively attributed to colorectal cancer, 33 to other causes, and 48 remained undetermined, a reflection of the practical difficulties of tracking very old patients who move in with family or enter nursing facilities. Yet every single one of the 18 confirmed cancer-related deaths occurred in patients with a GNRI below 98, and not one occurred in the nutritionally robust group. The authors caution that the large proportion of undetermined deaths makes cancer-specific survival analysis unreliable, but the pattern is difficult to dismiss entirely and hints that malnutrition and cancer lethality may travel together in this age group.

The mechanisms linking low GNRI to poor outcomes are biologically plausible. Malnutrition impairs wound healing, suppresses immune function, and increases susceptibility to infection, raising the risk of surgical site infections, anastomotic leakage, and pneumonia. It also drives muscle wasting and frailty, eroding the functional capacity that nonagenarians need to recover from a laparotomy or colectomy. Low albumin may additionally signal chronic systemic inflammation, which is independently associated with adverse surgical outcomes and tumor progression. In this framing, the GNRI is not just a number but a composite snapshot of the physiological terrain on which both the surgery and the cancer will play out.

The study has honest limitations. Its retrospective design and complete-case analysis excluded 177 of 402 eligible patients, and the excluded patients differed significantly in ASA class, surgical history, and cancer stage, so the analyzed cohort may not fully represent the source population. The GNRI itself cannot capture micronutrient deficiencies or inflammatory markers, and richer tools such as the Subjective Global Assessment or Comprehensive Geriatric Assessment might add depth. The cohort-derived cutoff of 86 showed only modest discriminatory ability, with an area under the curve of 0.593, and the authors explicitly label it exploratory and in need of external validation. Still, the clinical message is clear and actionable: every nonagenarian being considered for colorectal cancer surgery should undergo GNRI screening, and those scoring low should be referred for preoperative nutritional optimization, whether through dietary counseling, supplementation, or other interventions. As the world’s population ages and the oldest-old increasingly present with operable cancers, a two-variable formula that costs almost nothing to compute may become one of the most consequential numbers on the preoperative chart.

Subject of Research: Preoperative nutritional risk assessment using the Geriatric Nutritional Risk Index in nonagenarians undergoing colorectal cancer surgery

Article Title: The Geriatric Nutritional Risk Index Predicts Short‐ and Long‐Term Outcomes in the Oldest‐Old With Colorectal Cancer: A Multi‐Institutional Analysis of 225 Nonagenarians

Article References: Toshima, T., Teraishi, F., Takanaga, S., Mitsuhashi, T., Inada, R., Ohtani, T., Yoshida, R., Hori, N., Shigemitsu, K., Yamamoto, S., Kubota, T., Okano, Y., Nobuhisa, T., Taniguchi, F., Ishikawa, W., Shoji, R., Matsuda, T., Umeoka, T., Fujiwara, T., & The Setouchi Colorectal Neoplasm Registration Study Group Collaborators (2026). The Geriatric Nutritional Risk Index Predicts Short‐ and Long‐Term Outcomes in the Oldest‐Old With Colorectal Cancer: A Multi‐Institutional Analysis of 225 Nonagenarians. Annals of Gastroenterological Surgery, Article ags3.70288. https://doi.org/10.1002/ags3.70288

Image Credits: AI Generated

DOI: 10.1002/ags3.70288

Keywords: colorectal cancer, nonagenarians, Geriatric Nutritional Risk Index, nutritional status, surgical outcomes, overall survival, serum albumin, frailty, elderly surgery, propensity score analysis, Clavien-Dindo complications, preoperative assessment