Combination of lymphocyte count and albumin concentration as a new prognostic biomarker for rectal cancer

Although numerous studies have highlighted the prognostic values of various inflammation-related markers, clinical significance remains to be elucidated. The prognostic values of inflammation-related biomarkers for rectal cancer were investigated in this study. A total of 448 patients with stage II/III rectal cancer undergoing curative resection were enrolled from the discovery cohort (n = 240) and validation cohort (n = 208). We comprehensively compared the prognostic values of 11 inflammation-related markers-derived from neutrophil, lymphocyte, platelet, monocyte, albumin, and C-reactive protein for overall survival (OS) and recurrence-free survival (RFS). Among 11 inflammation-related markers, only “lymphocyte × albumin (LA)” was significantly associated with both OS and RFS in the discovery cohort (P = 0.007 and 0.015, respectively). Multivariate analysis indicated that low LA was significantly associated with poor OS (hazard ratio [HR] 2.19, 95% confidence interval [CI] 1.09–4.58, P = 0.025), and poor RFS (HR 1.61, 95% CI 1.01–2.80, P = 0.048). Furthermore, using the discovery cohort, we confirmed that low LA was significantly associated with poor OS (HR 2.89, 95% CI 1.42–6.00, P = 0.002), and poor RFS (HR 1.79, 95% CI 1.04–2.95, P = 0.034). LA can be a novel prognostic biomarker for stage II/III rectal cancer.

. A total of 240 patients comprised 163 males and 77 females, and their median age was 66 years old [range  were included. Tumor location was Ra in 122 patients (51%), and Rb-P in 118 patients (49%). As a preoperative treatment, NAC was administered in 46 patients (19%) and nCRT in 24 patients (10%). The 5-year OS and RFS of all the 240 patients were 87.4% and 72.6%, respectively.
The prognostic impacts of the 11 inflammation-related markers were investigated by the log-rank test with the Kaplan-Meier method ( Table 2). Univariate analysis indicated that low LA, the product of the lymphocyte count multiplied by the albumin concentration, was significantly associated with both poor OS and RFS (P = 0.007 and 0.015, respectively). The remaining combinations other than LA were not correlated with either OS or RFS in this cohort. Similarly, the 2 scoring systems (i.e., GPS and SIS) were not correlated with OS or RFS ( Supplementary Fig. S2).
During the study period, there were 37 deaths (15%) and 55 recurrences (23%). We further analyzed receiveroperating characteristic curves of the 9 combinations, and found that LA had the highest area under the curve values to predict OS and RFS (0.58 and 0.58, respectively) (Supplementary Figs. S3, S4).
Furthermore, univariate and multivariate analyses for RFS were conducted using clinicopathological characteristics and LA (Table 4, right). Multivariate analysis using Cox's proportional hazard regression model indicated that low LA remained to be significantly associated with poor RFS (HR 1.79, 95% CI 1.04-2.95, P = 0.034). Figure 2d shows the Kaplan-Meier curve of LA. Collectively, these results regarding the prognostic potential of LA in the validation cohort were almost consistent with those in the discovery cohort.
Stratification of the prognostic impact of LA levels. These aforementioned results indicated that LA was a robust biomarker for rectal cancer patients' prognosis. Therefore, we further investigated the prognostic value among the different LA levels. In the discovery cohort, the upper and lower quartiles of LA were 7920 and 4515, respectively. All 240 patients were classified into three groups based on the interquartile ranges of LA: (1) higher LA group (LA > 7920; n = 59), (2) middle LA group (4515 < LA ≤ 7920; n = 121), and (3) lower LA group (LA ≤ 4515; n = 60). Kaplan-Meier curves of the three groups on OS and RFS are shown in Fig. 3a,b. The lower LA group exhibited much poorer prognosis on OS and RFS than the higher LA group (P = 0.028 and P = 0.038, respectively).
Furthermore, we perform the same analysis in the validation cohort. All 208 patients were classified into three groups using the same cut-off values: (1) higher LA group (LA > 7920; n = 40), (2) middle LA group (4515 < LA ≤ 7920; n = 109), and (3)  www.nature.com/scientificreports/ on OS and RFS are shown in Fig. 3c,d. The lower LA group exhibited much poorer prognosis on OS and RFS than the higher LA group (P = 0.042 and P = 0.084, respectively) and middle LA group (P < 0.001 and P = 0.029, respectively). Taken together, these results indicated that LA could be a useful marker to predict the prognosis of rectal cancer.

Discussion
Although a growing number of studies have highlighted the prognostic values of various inflammation-related markers in several types of cancer [7][8][9][10][11][12][13][14][15][16][17][18][19] , clinical significance of inflammation-related markers remains to be elucidated. Moreover, there is a lack of consensus about the cut-off values for each marker. In this study, we focused on the patients with stage II/III rectal cancer, because accurate prediction of their prognosis can help decisionmaking in postoperative management, such as the follow-up schedule and need for adjuvant chemotherapy. In addition, the treatment strategy for stage IV rectal cancer is more complex compared with that for stage II or III, which indicates the need to investigate stage IV rectal cancer separately.
In the present study, we comprehensively evaluated the predictive potential for prognosis of 11 inflammationrelated biomarkers (i.e., NLR, PLR, LMR, PAR, MAR, NP, LA, NM, MP, GPS, and SIS) to identify the best combination of peripheral blood cell-based markers for stage II/III rectal cancer, and found that LA was a more reliable  www.nature.com/scientificreports/ prognostic predictor compared with any other inflammation-related markers in the discovery cohort (Tables 2,  3; Fig. 2a,b; Supplementary Fig. S3, S4). Of note, this prognostic impact of LA could be almost confirmed using the external validation cohort (Table 4; Fig. 2c,d). Although the potential causal effect behind the association of the LA with prognosis is not clear, it is possible to postulate the following hypotheses. Lymphocytes play a crucial role in the host cell-mediated cytotoxic immune response against tumors. A high density of tumor-infiltrating lymphocytes is strongly correlated with a favorable prognosis in several types of cancers 22 , suggesting that antitumor immune response is mediated mainly by lymphocytes. Serum albumin is produced from the liver, and is known as one of the negative acute phase proteins in response to inflammation. Furthermore, low albumin concentration also means malnutrition, and this can negatively affect tumor immunity in the microenvironment. Considering these findings, LA may reflect both the immunological response, represented by the lymphocyte count, and nutritional status, represented by the serum albumin level. Serum CRP level is a well-known established inflammatory marker. Combinations including the preoperative CRP level have been reported as prognostic markers in CRC. A recent study reported a significant potential of LCR, lymphocyte-CRP ratio. Okugawa et al. compared 9 inflammation-related markers in a discovery cohort of 373 patients with stage I-IV CRC, and found that LCR was a new reliable prognostic marker compared with Table 2. The relationships between inflammation-related markers (the 9 combinations and 2 scoring systems) and survival. NLR neutrophil-to-lymphocyte ratio, PLR platelet-to-lymphocyte ratio, PAR platelet-to-albumin ratio, LMR lymphocyte-to-monocyte ratio, MAR monocyte-to-albumin ratio, NP neutrophil × platelet, LA lymphocyte × albumin, NM neutrophil × monocyte, MP monocyte × platelet, GPS Glasgow prognostic scale, SIS systemic inflammation scale. * P < 0.05.  16 . In our institution, however, CRP levels lower than 0.1 mg/dL were judged as "zero"; thus, they could not be multiplied or divided by another variable. Furthermore, the distribution of serum CRP levels was too skewed (Table 1; Supplementary Fig. S1). Among 240 patients, 181 patients (75%) had CRP levels ≤ 0.2 mg/dL, that is, most of the study patients had the CRP levels within the normal range. Therefore, we assume that the variables incorporating the preoperative CRP level are difficult to become candidates for prognostic markers. Meanwhile, it is worth noting that the distribution of LA followed a normal distribution. Although we have recently reported the combination of GPS and NLR can be useful to predict the prognosis of CRC patients 14 , we could not identify the prognostic value of either NLR or GPS in this study (Table 2; Supplementary Fig. S2a,b). Recently, Suzuki et al. reported that SIS was a powerful predictor of OS and RFS in 727 patients with Stage I-IV CRC 17 . However, we could not identify the prognostic value of SIS in the discovery cohort ( Table 2, Supplementary Fig. S2c,d). These differences might be due to our focusing on rectal cancer or our exclusion of stage I and IV cancers.

5-year OS (%) P value 5-year RFS (%) P
This study has some limitations. First, this was a retrospective study with a relatively small population, and thus may be influenced by selection bias. Second, our treatment strategy was subject to certain chronological changes. Third, we included the patients with short observational period or censored cases shortly after operation, and this is a limitation to construct a prediction model for prognosis. In the future, we need to reevaluate the prediction performance of LA using ROC analysis or a deep learning method after a long period of observation 23,24 . Thus, further studies are desirable to establish the clinical utility of the newly developed LA. Table 3. Univariate and multivariate analysis using clinicopathological characteristics and LA for overall survival (OS) and recurrence-free survival (RFS) in the discovery cohort. CEA carcinoembryonic antigen, well/mod well-differentiated/moderately-differentiated, por/muc poorly-differentiated/mucinous, LA lymphocyte × albumin, HR hazard ratio, CI confidence interval. * P < 0.05. www.nature.com/scientificreports/ As far as we know, this is the first report to reveal the prognostic value of LA in rectal cancer. In stage II/III rectal cancer, assessment of preoperative LA can identify the high-risk subgroup for recurrence, which might help physicians to decide the postoperative management for preventing recurrence.
In conclusion, we found that "lymphocyte × albumin (LA)", the product of the lymphocyte count multiplied by the albumin concentration was a novel prognostic biomarker for stage II/III rectal cancer patients. This novel marker is easily calculated through routine blood tests, which can provide opportunities for further investigation. In all cases, the tumor was within 10 cm from the anal verge. Tumor location was defined according to the Japanese classification of Colorectal Carcinoma: "Ra", was the portion between S2 sacral vertebra and peritoneal reflection, "Rb" was the portion between peritoneal reflection and anal canal, and "P" was the portion within the anal canal 25 . Clinical stage was determined based on colonoscopy, magnetic resonance imaging (MRI), and computed tomography (CT). The institutional review board of Kyoto University approved the study protocol (reference No. R1958), and conformed to the Declaration of Helsinki. Informed consent was obtained from all participants.

Materials and methods
Treatment protocol. All surgeries were conducted by board-certified laparoscopic colorectal surgeons 26 .
Treatment strategies of individual cases were determined based on the multidisciplinary team meeting. Preoperative treatment (neoadjuvant chemoradiotherapy [nCRT] or neoadjuvant chemotherapy) was administered for patients with a high risk of recurrence, such as marked enlargement of regional lymph nodes, or bulky tumors (e.g., tumor size > approximately 50 mm, or clinically suspected involvement of circumferential resection margin) 27 . In the nCRT group, radiation therapy (total 45 Gy in 25 fractions) with concomitant CPT-11 and S-1 was administered. In the NAC group, FOLFIRI-based chemotherapy, FOLFOX6-based chemotherapy, or CPT-11 plus S-1 regimen was administered.
Follow-up investigation. The patient follow-up was standardized as follows; physical examination and blood test were done every 3 months in the first 3 years and every 6 months thereafter. CT was done every  www.nature.com/scientificreports/ 6 months (for stage II) or every 3 months (for stage III) in the first 3 years, and subsequently continued every 6 months for more than 2 years. Colonoscopy was done at 1, 3, and 5 years after operation. Recurrence was diagnosed by the imaging examinations and histopathological findings. OS was measured from the date of the first surgery to the date of death. RFS was measured from the date of the first surgery to the date of cancer recurrence or the date of death.
Inflammation-related markers. Blood samples were collected within one month before the surgery. In order to identify the inflammation-related prognostic markers, we chose 5 parameters available in a routine blood test: neutrophil, lymphocyte, platelet, monocyte, and albumin (Fig. 1). The 5 parameters were characterized as either up-regulating group (i.e., neutrophil, platelet, and monocyte) or down-regulating group (i.e., lymphocyte and albumin) 15,16,20,21 . Afterwards, we generated 9 combinations from the 5 parameters: NLR, PLR, LMR, LA, platelet-to-albumin ratio (PAR), monocyte-to-albumin ratio (MAR), neutrophil × platelet (NP), neutrophil × monocyte (NM), and monocyte × platelet (MP). When each of the combined parameters belonged to the same group, they were multiplied; when they belonged to different groups, they were divided. In addition, we chose GPS and SIS as the established inflammation-related scoring systems. Regarding GPS, patients with both higher CRP (> 1.0 mg/dL) and lower albumin concentration (< 3.5 g/dL) were categorized as GPS = 2, patients with either of these two were categorized as GPS = 1, and patients with neither were categorized as GPS = 0 11,19 . Regarding SIS, patients with both lower albumin concentration (< 4.0 g/dL) and lower LMR (< 4.44) were categorized as SIS = 2, patients with either of these two were categorized as SIS = 1, and patients with neither were categorized as SIS = 0 17,18 . These 11 inflammation-related markers composed of the 9 combinations and 2 scoring