Background

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive malignancies and has a 5-year survival rate of <10% [1]. Its incidence is continuing to increase, and it is now the third leading cause of cancer death worldwide [2]. Currently, the only potentially curative therapy is curative resection; however, few patients are eligible for this treatment. Therefore, the early detection of PDAC is critical to improving patient outcomes.

Serological markers have been widely used as first step for detection of cancers. Carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) are tumor markers that are commonly used in the diagnosis of PDAC [3]. However, they show limited usefulness for PDAC because of their relatively low sensitivity and specificity [4, 5]. Therefore, the identification of a more effective serum marker for PDAC is an urgent issue.

Laminin (Lm)-332 is a heterotrimeric basement membrane protein composed of Lm-α3, β3, and γ2 laminin chains. Basement membrane proteins are crucial for epithelial cell maintenance. We previously reported that Lm-γ2 is strongly expressed as a monomer in the tumor cell infiltrate in gastric adenocarcinoma but not in normal cells [6]. We further found that serum Lm-γ2 monomer (LG2m) may be a new diagnostic marker for hepatocellular carcinoma and bladder cancer [7, 8]. However, the role for LG2m as a biomarker in PDAC remains unclear. The aim of this study was to investigate the potential utility of LG2m as a diagnostic and prognostic marker in PDAC.

Methods

Patients

This study included consecutive patients who had been pathologically diagnosed with PDAC at the Kanazawa University Hospital (Cohort 1) and Kanagawa Cancer Center (Cohort 2) between October 2005 and December 2019. Individuals who were proven to be free of malignancy by medical examinations and patients who had intraductal papillary mucinous neoplasm (IPMN) diagnosed based on radiological findings without suggestive of high-grade features, i.e., “high-risk stigmata” or “worrisome features” as defined by international consensus guidelines for management of IPMN were included as negative controls for Cohort 1.

Detection of LG2m in sera by chemiluminescence immunoassay (CLIA)

Anti-LG2m monoclonal antibody (2H2) was immobilized on magnetic microparticles using carboxylic acid groups, and the anti-LG2m DIII domain mouse monoclonal antibody was labeled using acridinium. The fully automated CLIA performed a two-step sandwich assay using the ARCHITECT system (Abbott Laboratories, Chicago, IL, USA) (https://doi.org/10.1186/s40364-017-0109-4). The recombinant LG2m protein was diluted in phosphate buffered saline (PBS) containing 1% (w/v) BSA, 0.1% (v / v) Tween-20 and 0.1% (v / v) ProClin300 and was used to develop the assay standards. The standard concentrations of LG2m were 0, 20, 50, 100, 1000, and 20,000 pg/mL.

Data collection

The medical records and collected clinicopathological data of all patients were reviewed. The clinicopathological data included information on age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS), laboratory data, tumor markers, pathological diagnosis, imaging data, Unio Internationalis Contra Cancrum TNM stage, and patient outcome. The Institutional Review Boards of Kanazawa University Hospital and of Kanagawa Cancer Center approved the study protocol (IRB number: 2016-098 and 2018eki-52, respectively), and the study was conducted in accordance with the Declaration of Helsinki. Informed consent from the patients obtained for all medical procedures, participation to this study and publication of this report.

Statistical analysis

The optimal cut-off level of LG2m as a diagnostic and prognostic marker for PDAC has not been determined. Therefore, we identified the cut-off level in this study by receiver operating characteristic analysis through the comparison between healthy individuals and PDAC patients and between IPMN patients and PDAC patients. The institutional criteria for the CEA and CA19-9 cut-off levels are 5.0 ng/mL and 39 U/mL, respectively; however, the cut-off levels for predicting patient prognosis have been unclear. Therefore, we used median levels as the cut-off levels for CEA, CA19-9 and LG2m. The chi-squared test was used to evaluate relationships between groups. Overall survival (OS) was calculated from the date of diagnosis to the date of death or the last day of the follow-up period. To compare OS between groups, cumulative survival was calculated using the Kaplan–Meier method and differences between groups were evaluated using the Cox’s proportional hazards regression model. A P-value of less than 0.05 indicated statistical significance. Statistical analyses were performed using Stata 12.1 (College Station, TX, USA).

Results

Subject

A total of 243 patients were in Cohort 1, including 142 patients with pathologically proven PDAC, 55 patients clinically diagnosed with IPMN, and 46 healthy individuals. Cohort 2 included 518 patients clinically diagnosed with PDAC at Kanagawa Cancer Center. The patient characteristics are summarized in Table 1.

Table 1 Patient characteristics.
Full size table

Comparison of serum LG2m levels between patients with PDAC, patients with IPMN, and healthy individuals

Serum LG2m levels were successfully evaluated from all samples in Cohort 1, and the median serum LG2m level of all samples was 14.2 pg/mL The median serum LG2m levels in patients with PDAC, patients with IPMN, and healthy individuals in Cohort 1 were 22.5 pg/mL, 11.0 pg/mL, and 4.9 pg/mL, respectively (Fig. 1a). The serum LG2m level of patients with PDAC was significantly higher than that of healthy individuals (P < 0.001) and patients with IPMN (P < 0.001).

Fig. 1: Box plots showing serum LG2m levels.
Fig. 1: Box plots showing serum LG2m levels.
Full size image

a The median serum LG2m levels in healthy individuals, IPMN patients, and PDAC patients were 4.9, 11.0, and 22.5 pg/mL in Cohort 1. The serum LG2m level of patients with PDAC was significantly higher than that of healthy individuals (P < 0.01) and that o patients with IPMN (P < 0.01), respectively. b The median serum LG2m levels in PDAC patients was 26.1 pg/mL in Cohort 2.

In the comparison between PDAC patients and healthy individuals, receiver operating characteristic analysis identified the optimal cut-off level of LG2m as 9.55 pg/mL. With 9.55 as the cut-off level of LG2m, the sensitivity, specificity, and area under the curve were 0.89, 0.87, and 0.88, respectively (Fig. 2a). The maximum LG2m level was 20.1 pg/mL in the 46 healthy individuals. Using 20.1 as the cut-off level, the specificity was 1 and the sensitivity was below 59.2%.

Fig. 2: Receiver operating characteristic curves of the results comparing PDAC patients and healthy individuals and IPMN patients.
Fig. 2: Receiver operating characteristic curves of the results comparing PDAC patients and healthy individuals and IPMN patients.
Full size image

a The sensitivity, specificity and area under the curve were 0.89, 0.87 and 0.88, respectively, when the cut-off level of LG2m was set as 9.55 in comparing PDAC patients and healthy individuals. b The sensitivity, specificity and area under the curve were 0.77, 0.80 and 0.79, respectively, when the cut-off level of LG2m was set as 14.03 in comparing PDAC patients and IPMN patients.

In the comparison between PDAC patients and IPMN patients, receiver operating characteristic analysis identified the optimal cut-off level of LG2m as 14.03 pg/mL. Using 14.03 as the cut-off level of LG2m, the sensitivity, specificity, and area under the curve were 0.77, 0.80, and 0.79, respectively (Fig. 2b).

Correlation between serum CEA and CA19-9 levels and serum LG2m level

In Cohort 1, the median serum CEA and CA19-9 levels were 4.1 ng/mL and 321.3 U/mL among PDAC patients, 2.3 ng/mL and 5.8 U/mL in IPMN patients, and 1.6 ng/mL and 4.4 U/mL in healthy individuals, respectively. The sensitivity of CEA and CA19-9 for PDAC was 0.41 and 0.77, and the specificity of CEA and CA19-9 for healthy individuals was 0.98 and 0.98, respectively.

The correlation between serum CEA and LG2m level is shown in Fig. 3a. No significant correlation was found between serum CEA and LG2m levels (r = 0.036, P = 0.58). The correlation between serum CA19-9 and LG2m level is shown in Fig. 3b. No significant correlation was found between serum CA19-9 and LG2m levels (r = 0.027, P = 0.68).

Fig. 3: Correlation between serum tumor markers.
Fig. 3: Correlation between serum tumor markers.
Full size image

a No significant correlation was found between serum CEA and LG2m levels. b No significant correlation was found between serum CA19-9 and LG2m levels. c The prevalence of patients with elevated serum CEA, CA19-9, and LG2m in Cohort 1. d The prevalence of patients with elevated serum CEA, CA19-9, and LG2m in Cohort 2.

Among the 84 PDAC patients with serum CEA level lower than 5.0 ng/mL, the median serum LG2m level was 21.1 pg/mL; the serum LG2m level was more than 9.55 pg/mL in 72 patients (85.7%). Among the 16 PDAC patients with serum LG2m level lower than 9.55 pg/mL, the median serum CEA level was 2.65 and the serum CEA level was more than 5.0 ng/mL in 4 patients (25.0%) (Fig. 3c). Among the 32 PDAC patients with serum CA19-9 level lower than 39 U/mL, the median serum LG2m level was 20.3 pg/mL and the serum LG2m level was more than 9.55 pg/mL in 26 patients (81.3%). Among the 16 PDAC patients with serum LG2m level lower than 9.55 pg/mL, the median serum CA19-9 level was 129.5 U/mL and serum CA19-9 level was more than 39 U/mL in 10 patients (62.5%) (Fig. 3c). All but five patients had elevated levels of at least one of LG2m, CA19-9 or CEA.

External validation of serum LG2m in PDAC patients

The median serum LG2m level of all 518 patients with PDAC in Cohort 2 was 26.1 pg/mL, and the serum LG2m level was more than 9.55 pg/mL in 452 patients (87.3%); these results were consistent with those of Cohort 1 (Fig. 1b). Age, sex, and tumor location did not affect serum LG2m level, but more patients with advanced stage disease had serum LG2m levels of over 9.55 pg/mL (Supplementary Table). The median levels of serum LG2m in patients with UICC stage I, II, III, and IV were 15.8 pg/mL, 21.2 pg/mL, 23.0 pg/mL, and 38.7 pg/mL, respectively (Fig. 4). Stage I patients had lower LG2m levels than patients with more advanced stages; however, 31 of the 43 patients with stage I disease (72.1%) had LG2m levels higher than 9.55 pg/mL (Supplementary Table). Moreover, 157 of the 190 (80.5%) patients with stage II disease, which is eligible for curative resection, had LG2m levels higher than 9.55 pg/mL.

Fig. 4: Box plots showing serum LG2m levels of patients with PDAC by UICC stage in Cohort 2.
Fig. 4: Box plots showing serum LG2m levels of patients with PDAC by UICC stage in Cohort 2.
Full size image

The median serum LG2m levels in PDAC patients among UICC stage I, II, III, and IV were 15.8, 21.2, 23.0, and 38.7 pg/mL in Cohort 2. Stage I patients had lower LG2m levels than those with more advanced stages, but 31 of 43 patients (72.1%) had LG2m levels higher than 9.55 pg/mL, which level were set as optimal cut-off level determined by receiver operating characteristic analysis in Cohort1.

The median serum CEA and CA19-9 levels were 4.1 ng/mL and 475.7 U/mL, respectively, in Cohort 2. The sensitivities of CEA and CA19-9 were 41.3% and 77.4%, respectively; these results were consistent with those of Cohort 1. The sensitivity of CEA in patients with stage I and stage II PDAC was 9.3% and 24.7%, which was significantly lower than that of LG2m (P < 0.001 and P < 0.001, respectively). The sensitivity of CA19-9 in patients with stage I and stage II PDAC was 53.5% and 74.7%, which tended to be lower than that of LG2m (P = 0.074 and P = 0.18, respectively), although there was no significant difference.

In Cohort 2, among the 304 PDAC patients with serum CEA level lower than 5.0 ng/mL, 249 patients (81.9%) had serum LG2m level higher than 9.55 pg/mL. Similarly, among the 117 PDAC patients with CA19-9 level lower than 39.0 U/mL, 103 patients (88.0%) had serum LG2m level higher than 9.55 pg/mL (Fig. 3d).

Patient outcome by serum LG2m level

In Cohort 2, resection, chemo-radiotherapy, and chemotherapy were selected as initial treatment in 218 (42.1%), 7 (13.5%), and 275 patients (53.1%), respectively; the remaining 18 patients (3.5%) were treated with supportive care (Table 1). Combination therapy with gemcitabine and albumin-bound paclitaxel (nab-paclitaxel) was most commonly used first-line chemotherapy regimen in 224 patients.

In the patients who received resection as initial treatment for PDAC, the median duration of follow-up was 23.3 months (1.68–114.4), and 107 patients (49.1%) died by the day of data collection cut-off. The median OS of all patients was 32.4 months, and the 1-year, 2-year and 3-year survival rates were 84.8%, 62.8%, and 46.9%, respectively. Among the patients who received resection, patients with serum LG2m level higher and lower than 18.9 pg/mL had an OS of 26.9 months and 39.3 months, respectively; patients with serum LG2m levels higher than median levels had a significantly shorter survival than those with serum LG2m level lower than the median level (P = 0.031) (Fig. 5a). In the patients who received resection, pretreatment high serum LG2m level was one of the independent unfavorable factors for OS (hazard ratio (HR), 1.536; P = 0.042); other unfavorable factors included CA19-9 140.0 U/mL or more (HR 2.024; P < 0.001), age 66–73 years old (HR compared with 74 years old or older, 2.430; P = 0.001), age 65 years old or younger (HR compared with 74 years old or older, 1.798; P = 0.027), and absence of adjuvant therapy after resection (HR 6.591; P < 0.001) (Table 2).

Fig. 5: Patient outcome by serum LG2m levels in Cohort 2.
Fig. 5: Patient outcome by serum LG2m levels in Cohort 2.
Full size image

Patient outcome by serum LG2m levels in patients who received a resection and b chemotherapy using gemcitabine and albumin-bound paclitaxel in Cohort 2. Patients with higher serum LG2m level than median level had significantly shorter survivals than those with lower serum LG2m level than median level.

Table 2 Prognostic factors for overall survival in patients who received resection.
Full size table

In the patients who received gemcitabine plus nab-paclitaxel as initial treatment for PDAC, the median duration of follow-up was 11.2 months (0.63–57.8), and 162 patients (72.3%) died by the day of data collection cut-off. The median OS of all patients was 12.9 months, and the 6-month, 1-year, and 2-year survival rates were 78.9%, 52.7%, and 20.3%, respectively. The median OS of patients with serum LG2m level higher and lower than 34.9 pg/mL was 8.3 months and 16.4 months, respectively. Patients with serum LG2m levels higher than the median level had a significantly shorter survival than those with serum LG2m level lower than the median level (P < 0.001) (Fig. 5b). Pretreatment high serum LG2m level (HR, 1.781; P < 0.001) and UICC stage IV (HR 3.061; P = 0.003) were the independent unfavorable factors for OS in the patients who received gemcitabine plus nab-paclitaxel (Table 3). Patients with CA19-9 higher than the median level tended to have a worse OS than the patients with CA19-9 lower than the median level, but the difference was not significant (P = 0.087) in a multi-variate Cox regression model.

Table 3 Prognostic factors for overall survival in patients who received chemotherapy with gemcitabine and albumin-bound paclitaxel.
Full size table

Discussion

In this study, we investigated the usefulness of serum LG2m as a novel diagnostic and prognostic marker for PDAC. First, we found that the serum LG2m levels were significantly higher in PDAC patients compared with healthy individuals, and both the sensitivity and specificity were approximately 0.9, which was very high, indicating the usefulness for serum LG2m level in the diagnosis of PDAC. We confirmed the high sensitivity of serum LG2m level in Cohort 2, the other large PDAC cohort. We also showed the usefulness of serum LG2m as a prognostic marker in patients who received resection and chemotherapy with gemcitabine plus nab-paclitaxel.

The main reason that pancreatic cancer has the most dismal prognosis among carcinomas is that early detection is difficult, and at the time of diagnosis, most patients are already in a condition where they cannot undergo resection, the only curative treatment [9]. Despite the advancement of diagnostic techniques, the inability to efficiently identify high-risk populations of PDAC has led to the need for simpler surveillance methods, and one of the solutions is sensitive serological markers [10]. CA19-9 and CEA are serological markers of PDAC; however, their sensitivity is not high, especially in the early stage [11, 12]. Our results suggested that LG2m may play an important role in the screening or surveillance of PDAC as a more sensitive and easily measured serological marker.

Notably, the present study showed that LG2m did not correlate with CA19-9 or CEA, as shown in Fig. 3a, b. PDAC with elevated CA19-9 or CEA and PDAC with elevated LG2m may have different biological characteristics; however, this study was unable to characterize PDAC with elevated LG2m other than more advanced stage. In Cohort 1, the sensitivity of LG2m was 88.7%, which is notably high, and the sensitivity was further increased to 95.8% by adding CA19-9. Similarly, in Cohort 2, the sensitivity of LG2m increased from 87.3% to 97.3% when LG2m was combined with CA19-9. We believe that the near 100% sensitivity by the combined use of these tumor markers is of high diagnostic value considering the position of these serological markers as screeners.

To apply LG2m for the detection of PDAC in the general population, in the differentiation from benign disease such as IPMN, or for the prediction of prognosis of patients who received curative resection or unresectable cases after diagnosis of PDAC, it is necessary to establish the appropriate cut-off of LG2m for each population. In this study, sensitivity and specificity were calculated using a cut-off level of 9.55 pg/mL from the receiver operating characteristic curve for the results comparing PDAC patients and healthy individuals; however, a lower cut-off level may be better for detection of PDAC in an earlier stage. The most appropriate cut-off level for prognosis prediction, even for CA19-9, has not been determined. In this study, we used the median levels and set the cut-off as 18.9 pg/mL for LG2m and 140.0 U/mL for CA19-9 for the patients that were treated with resection in Cohort 2 and 34.9 pg/mL for LG2m and 1155 U/mL for CA19-9 for the patients treated with chemotherapy in Cohort 3. Determining the precise optimal cut-offs may require further study.

The results of Cohort 2 confirm the high sensitivity of this new diagnostic marker was comparable to that of Cohort 1. If negative controls could be included in the validation cohort as in Cohort 1, we could be able to validate not only the sensitivity but also more parameters such as specificity and AUC. Unfortunately, we were not able to include healthy subjects or patients with benign disease at Kanagawa Cancer Center. Therefore, this point also needs to be verified in a large cohort.

In conclusion, serum LG2m levels were higher in patients with PDAC than in healthy individuals, which suggested that LG2m may be a useful diagnostic marker for PDAC. Additionally, LG2m may be useful in predicting the prognosis of PDAC patients after resection or chemotherapy with gemcitabine plus nab-paclitaxel, suggesting that LG2m combined with CA19-9 may contribute to improving the prognosis of PDAC patients. We are currently conducting a prospective study to confirm the findings of this study.

Supplementary information is available at the British Journal of Cancer’s website.