Gastric cancer (GC) is one of the most frequently diagnosed malignancies, with both the incidence and mortality ranked fifth worldwide (1). Unfortunately, nearly 70% of GC patients in Western countries and China are diagnosed at the locally advanced stage and even at the advanced stage (2–4). At the present time, curative resection (radical gastrectomy and D2 lymphadenectomy) and peri-operative chemotherapy is the recommended standard management for locally advanced GC (LAGC) (5–7). Although GC managements have dramatically improved, the prognosis of LAGC remains unsatisfactory. Recurrences commonly occur the majority of which appear within 2 years after surgery (8). The pathological tumor-lymph node-metastasis (TNM) classification system is the most frequently utilized predictor for prognosis of patients with GC. But in clinical practice, it is fairly common to encounter patients with the same tumor stage but with significantly different long term outcomes. Therefore, it is necessary to investigate further the underlying mechanism(s) of the disease and other influencing factors (9, 10).

There is increasing evidence that a patient’s immune and nutrition status is associated with not only post-operative complications, but also oncological outcomes of patients with various cancers (2, 9, 11–14). Among these, for example, serum albumin and prealbumin concentrations, neutrophil and lymphocyte counts and C reactive protein (CRP), are the most frequently utilized indicators for immune and nutrition status (15–18). In addition, several combined indexes, including the modified Glasgow Prognostic Score (mGPS) and prognostic nutritional index (PNI), have also been proposed and identified to be independent predictors for the prognosis of several types of malignancies (9, 11, 19). In one of our previous studies that involved 731 stage II or III GC patients who undergoing radical gastrectomy, the optimal cut-off values of prealbumin concentration and lymphocyte count for overall survival (OS) were set as 180 mg/L and 1.5 ×10⁹/L determined by X-tile (18). Further univariate analysis and multivariate Cox regression analyses confirmed that both a prealbumin concentration < 180 mg/L and a lymphocyte count < 1.5 ×10⁹/L were confirmed to be significant predictor for disease free survival (DFS) and OS. Thereafter, a new index, the combined prealbumin and lymphocyte (Co-PaL) score, was established by combining the prealbumin and lymphocyte values. The exact calculation method was as follows: patients with a prealbumin concentration < 180 mg/L and a lymphocyte count < 1.5 ×10⁹/L were given a Co-PaL score of 2, indicating severe malnutrition. Patients with one of these conditions were given a Co-PaL score of 1, indicating mild to moderate malnutrition. Patients with neither of these conditions were given a Co-PaL score of 0, indicating good nutrition. Internal validation (subgroup analyses) found that both OS and DFS were significantly different among patients with a Co-PaL score of 0, 1 and 2, separately. But these conclusions have not been unequivocally validated externally. In this retrospective cohort study involving a large sample size from a tertiary center, a nomogram was established to validate the predictive value of the Co-PaL score in stage II or III GC patients after radical gastrectomy.

In our previous study (18), both a lymphocyte count < 1.5 ×10⁹/L and a prealbumin concentration < 180 mg/L were identified as independent unfavorable predictors for OS of stage II/III GC patients following curative resection. Thereafter, the newly established Co-PaL score, could accurately classify patients into poor, mild to moderate, and good nutritional statuses and was a potential predictor for prognosis. In order to validate these findings, we conducted this retrospective study by analyzing the medical data of 890 consecutive patients with stage II or III GC. Except for the well-known tumor stage, post-operative complication and AC, the Co-PaL score was also identified as a significant risk factor for OS. Subgroup analyses found that regardless of the tumor stage, complications and AC, the Co-PaL score was significantly associated with prognosis. Furthermore, the established nomogram based on the above mentioned 4 predictors showed a good ability to predict the oncological outcomes. The calibration curves indicated satisfactory consistency between the actual and predicted survival probabilities at 1-, 3- and 5-years. Finally, DCA of the nomogram showed superiority compared to the TNM classification system. As listed in Supplementary Table 2, the basic characteristics between the development and validation cohort were significantly different, including age, BMI, prealbumin and albumin concentrations, the tumor stage and complications. Validation through a new dataset with a significant difference of basic characteristics ensured the promising predictive value of the Co-PaL score for prognosis in stage II/III GC patients who underwent gastrectomy.

The TNM stage was the most well-established predictor for treatment strategy and prognosis of GC (5–7, 23). As shown in Figure 2, the established nomogram revealed that stage III disease was 100 points higher compared to stage II disease, which was obviously higher than the other 3 variables. Thus, the pTNM stage was identified as the most significant predictor for prognosis in the present study. Post-operative complications, especially infections, are also well-known adverse indicators for the prognosis of various cancer types (2, 24–27). Possible explanations include that post-operative complications might delay AC, adversely impact the compliance of AC, cause systemic inflammation and weaken the hosts’ immunity against cancer cells (2, 24–28). To reduce and/or delay recurrence after resection, AC has been recommended as the standard management for LAGC worldwide (5–7). Consistent with previous study findings, AC has been confirmed as a protective factor for long-term survival in the present study (22, 27). But in clinical practice, it is not uncommon to come across patients who refuse to undergo AC, or could not finish the planned AC regimens for various reasons, such as delayed recovery from surgery, frailty and AC induced toxicity, and so on. Even in the prospective phase 3 CRITICS study, about a quarter of patients did not receive any chemotherapy or chemo-radiotherapy postoperatively as planned. In addition, another 25% of patients did not complete the planned chemotherapy regimen (29).

Several studies have found that the completeness of AC affected the prognosis of LAGC (30, 31). In the well-known CLASSIC study (30), post hoc analysis found that patients receiving 6 or more cycles of AC had a significantly better prognosis than those receiving less than 6 cycles. Our previous studies found that poorer nutritional statuses, which were assessed by PNI or BMI, independently affected the completeness of AC (31, 32). In the present study, patients with higher Co-PaL scores, were found to be less likely to receive AC (80.13%, 70.03% and 64.44% in those with Co-PaL score of 0, 1 and 2, respectively, P < 0.001). In addition, 45.03%, 42.65% and 37.78% of patients with Co-PaL scores of 0, 1 or 2 completed at least 6 cycles of peri-operative chemotherapy (P = 0.002). Thus, the Co-PaL score, which indicate nutritional and immune statuses, might serve as an indicator for compliance of AC, a suggestion which requires further research. Another previous study proved that oral nutritional supplements could improve nutritional outcomes and the completion of planned therapy (33). As a result, it is reasonable to suppose that such strategies might improve the long term outcomes of LAGC, but further prospective studies are required to verify this conjecture.

Malnutrition occurs commonly in cancer patients, especially in those with digestive system cancers, such as GC (34). The unique anatomical structure of GC results in an imbalance of nutrient intake and bodily requirements which might partly explain the underlying reasons. Malnutrition, in turn, adversely impacts immune functions. For example, malnutrition may result in an immune-suppressive micro-environment and increase the rate of invasive tumor formation (35). Nowadays, more and more evidence has accumulated showing that malnutrition and the impaired immune status is not only related to increased post-operative complications and prolonged duration of hospital stays, but also to increased chemotherapy-induced toxicity, influenced the compliance of chemotherapy, and as a result led to a poorer prognosis (9, 11, 14, 15, 31). Albumin, prealbumin, lymphocyte and CRP are the most commonly utilized indicators for immunological and nutritional statuses (15–18). But the conclusions of researchers have not always been uniquely consistent. In a retrospective study of 4,732 GC patients, a lower prealbumin concentration (< 15 mg/dL and 15-22 mg/dL vs. ≥ 22 mg/dL) was identified as significant risk factors for OS. Further analysis found that the prealbumin concentration was associated with other cause survival, instead of cancer specific survival (16). In one of our previous studies involving 731 stage II/III GC patients who underwent operations, a lower prealbumin concentration (< 180 mg/L), instead of albumin, was confirmed as an independent poor predictor for both OS and DFS (18). The possible explanation was that prealbumin has a significantly smaller body pool and shorter half-life than albumin. Thus prealbumin might serve as a more sensitive indicator for nutritional status.

On the other hand, the lymphocyte count is a simple and widely used indicator to assess immunity. Decreased lymphocyte numbers generally represents an immunosuppressed condition to the attack and elimination of cancer cells, resulting in an unsatisfactory prognosis of various cancer types (36, 37). In one of our previous research projects, 390 GC patients were enrolled from 3 tertiary hospitals who underwent radical gastrectomy and whose peripheral lymphocyte subsets were examined. A T lymphocyte count < 0.84 × 10⁹/L was confirmed to be a significant adverse predictor both in the development and validation groups. The ability of antigen directed cyto-toxicity of the T lymphocyte, which plays an essential role in cellular adaptive immunity fighting cancers, might partly explain our findings (38). But further prospective studies are still awaited to confirm these findings and explore the underlying mechanisms.

There were a number of limitations to the present study. First, it was retrospective in nature and conducted in a single-center; thus, selective bias was likely inevitable. Second, only about 10% of patients (90/890) were assigned a Co-PaL score of 2, which represented severe malnutrition in the present study. The relatively small numbers of patients in some subgroups might have had an impact on the robustness of our conclusions. Third, although the majority of relapses of LAGC occurred within 36 months following resection (39), the relatively short follow-up duration of 38 months may have been inadequate to gather and analyze late relapses and deaths. Lastly, but by no means least, external validation with large sample size from geographically/institutionally distinct centers is still required to verify our findings, especially in those patients from Western countries, given that management strategies for LAGC differ significantly among the Western and Eastern countries.

In conclusion, this study confirmed that the lymphocyte and prealbumin based Co-PaL score was a promising prognosticator in stage II or III GC after radical gastrectomy. The established nomogram incorporating the Co-PaL score exhibited a satisfactory predictive ability to predict the survival probability at 1-, 3- and 5-years. Further external validation is still needed to assess the accuracy of this nomogram.