- Email: [email protected]
Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients
European Journal of Surgical Oncology xxx (2018) 1e7
Contents lists available at ScienceDirect
European Journal of Surgical Oncology journal homepage: www.ejso.com
Consensus Statement
Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients Peng-liang Wang, Jin-yu Huang, Zhi Zhu, Bao-cheng Gong, Han-wei Huang, Shi-jie Duan, Hui-mian Xu**, Fu-nan Liu* Department of Surgical Oncology, The First Affiliated Hospital of China Medical University, Shenyang, China
a r t i c l e i n f o
a b s t r a c t
Article history: Accepted 30 January 2018 Available online xxx
Background: The status of serosal invasion is often discordance between pathological and intraoperative evaluation. Our study sought to develop a risk-scoring system (RSS) to predict the probability of pT4a for macroscopic serosal invasion (MSI) positive patients and reevaluate the serosal invasion status. Patients and Methods: A total of 1301 pT3/pT4a gastric cancer patients with curative surgery were reviewed. We constructed the RSS to predict the probability of pT4a and assigned MSI-positive patients into different risk groups based on the risk scores. The prognostic significance of these risk groups was also evaluated. Results: Univariate and multivariate analyses identified that tumor location, Lauren type, Borrmann type, tumor size, lymphovascular invasion and pN stage were risk factors related to pT4a. Survival analyses showed that pT3 MSI-positive patients in high-risk group had similar survival with pT4a patients. We incorporated these two groups into one stage and proposed a novel revised-T stage. Two-step multivariate analyses indicated that the revised-T stage showed better prediction ability for prognosis and peritoneal recurrence assessment than original pT stage and MSI status. Conclusions: In our present study, we developed a RSS to predict the probability of pT4a for MSI-positive patients. Based on our RSS, we proposed a treatment algorithm to reevaluate the tumor invasion for MSIpositive patients in clinical practice. Future studies should include other preoperative predictors to improve the clinical utility of our model. © 2018 Elsevier Ltd, BASO ~ The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved.
Keywords: Gastric cancer Macroscopic serosal invasion Risk-scoring system Peritoneal recurrence
Introduction Recently, American Joint Committee on Cancer/International Union Against Cancer (AJCC/UICC) tumor-node-metastasis (TNM) classification, which based on the depth of tumor invasion, number of regional metastasis lymph nodes and distant metastasis, is the most widely used staging of gastric cancer (GC) [1]. However, the pathological T stage is often inconsistent with the T stage evaluated by surgeons during operations, especially the status of serosal
* Corresponding author. Department of Surgical Oncology, The First Affiliated Hospital of China Medical University, 155 Nanjing North Street, Heping District, Shenyang, 110001, China. ** Corresponding author. Department of Surgical Oncology, The First Affiliated Hospital of China Medical University, 155 Nanjing North Street, Heping District, Shenyang, 110001, China. E-mail addresses: [email protected] (H.-m. Xu), [email protected] (F.-n. Liu).
invasion [2e4]. And about half of the patients were macroscopic serosal invasion (MSI) positive and finally identified as the serosallimited tumor invasion pathologically [4e6]. Several studies also demonstrated that MSI-positive patients had worse prognosis as well as higher incidence of peritoneal recurrence [3,5e8]. Peritoneal recurrence is the most common pattern of recurrence and almost half of advanced GC patients died of peritoneal recurrence even after curative resection [9e11]. Previous studies indicated that intraperitoneal free cancer cells are significantly associated with peritoneal recurrence [12e15]. When tumor invades into the serosal layer, cancer cells may exfoliate from the lesion where the normal serosal are disrupted by tumor invasion [8,16]. Thus, the status of serosal invasion is extremely critical for detection of intraperitoneal free cancer cells and prediction of peritoneal recurrence. Although hyperthermic intraperitoneal chemotherapy (HIPC) may improve the overall survival (OS) and decrease the peritoneal
https://doi.org/10.1016/j.ejso.2018.01.240 0748-7983/© 2018 Elsevier Ltd, BASO ~ The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
2
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
recurrence among patients with pathological serosal invasion [17], the efficacy of this treatment for MSI-positive patients remains controversial [18,19]. Considering the morbidity and mortality rate of this treatment, the accurate assessment of tumor invasion is prerequisite for these patients [20]. Based on above considerations, we sought to reevaluate the tumor invasion status for these patients by developing a riskscoring system (RSS).
evaluated by plotting the calibration curve and testing the statistical significance of Hosmer-Lemeshow goodness-of-fit test (P > 0.05 indicating good fit). Both of the c-statistic and the calibration curve were internally validated by bootstrap method to reduce the over fit bias (200 repetitions). Decision curve analysis (DCA) was constructed to evaluate the clinical usefulness of RSS [24]. This analysis estimates the clinical net benefit of our model by summing the benefits (true-positive) and subtracting the harms (false-positive).
Patients and Methods Survival analysis and peritoneal recurrence assessment Study population A total of 2062 GC patients who underwent gastric resection and combined D2/D3 lymphadenectomy and achieved R0 resection at the Department of Surgical Oncology, the First Affiliated Hospital of China Medical University from January 1987 to December 2006 were collected. Of these patients, 1366 patients were confirmed as pT3/pT4a pathologically according to the seventh edition of the UICC/AJCC TNM staging system [1]. We mainly included these patients for analyses because the majority of MSI-positive patients were finally identified as pT3/pT4a histologically [2,4,6,21]. Among 1366 patients, 28 patients died in the postoperative period (<3 months). All patients received follow-up after being discharged from hospital, every quarter of the first 3-year, 6 months during the fourth and fifth year, and once a year thereafter. Followup evaluations included patient history, physical examination, carcinoembryonic antigen (CEA), carbohydrate antigen19-9 (CA199) and CA72-4. The gastroscopy and computed tomography (CT) scan were performed once a year. At the time of the last follow-up (March 31, 2011), the median and mean follow-up time was 32.0 and 61.8 months, respectively (range 3e313 months). 24 patients were lost and these patients were excluded from our study (followup rate: 98.2%). Of the remaining 1314 patients, 13 patients with incomplete information were also excluded. Finally, 1301 patients were selected for further analysis. A total of 753 patients received postoperative intravenous adjuvant chemotherapy and/or intraperitoneal chemotherapy. No one received neoadjuvant chemotherapy and/or radiotherapy. At the time of the last follow-up, 622 patients died of cancer-related diseases and 362 patients died of other diseases or causes unknown. Peritoneal recurrence was diagnosed mainly by CT scan and sometimes identified by cytology of ascites or intraoperative biopsy if operations were performed. Of the 622 patients, 293 were identified with peritoneal recurrence. This study was approved by the Institutional Ethics Committee of China Medical University. Development and evaluation of risk-scoring system Logistic regression analysis was used to identify the risk factors of tumor invasion diagnosed as pT4a. The score of each risk factor was estimated based on the results of multivariate analysis. The bcoefficients of multivariate analysis were rounded up to the nearest half integer and then multiplied by 2 to avoid decimals [22]. We calculated the total points of each patient and divided the patients into quartiles approximately. Then three risk groups (low-, medium-, and high-risk group) were established with the middle two quarters combined. The total risk score of each patient was used as a risk factor for the model evaluation. The performance of the developed RSS was evaluated with respect to discrimination and calibration. Discrimination was measured by the concordance statistic (c-statistic), which is equivalent to the area under the receiver operating characteristic (ROC) curve [23]. The calibration of the model was
To elucidate the prognosis influence of our RSS, we compared the OS between these risk groups in each pN stage and pT stage. Meanwhile, we compared the OS of pT3 MSI-positive patients in different risk groups with pT4a MSI-positive patients. Based on these results, patients who had similar survival were incorporated into one T stage and a novel revised-T stage was formed. Additionally, two-step multivariate analysis was also used to compare the novel revised-T stage with original UICC/AJCC pT stage in the prediction of peritoneal recurrence and the prognosis assessment. Patients died of other diseases or with causes unknown (n ¼ 362) were excluded when we evaluated the peritoneal recurrence. Statistical analysis Continuous variates were described as mean ± standard deviation (SD) and categorical variates were described as count and percentages. Comparisons between groups were performed by using Pearson c2 test for categorical variates and two-tailed Student's t-test for continuous variates. Univariate analyses were performed to indentify the significant risk factors related with pT4a and peritoneal recurrence. Multivariate analyses were used to identify the independent risk factors. The OS rates were estimated by Kaplan-Meier curves and the comparisons were identified by the log-rank test. Multivariate Cox proportional hazards regression model was used to identify the independent prognosis predictors. Only those variates with P values < 0.05 in univariate analyses were selected into the multivariate analyses. The analyses were performed using the R software version 3.2.0 (http://www.r-project.org/) and SPSS version 21.0 (SPSS, Chicago, IL). A two-tailed P value < 0.05 was considered as statistically significant in all analyses. Results Study population and model development Clinicopathological characteristics of MSI-positive and -negative patients were presented in Supplementary Table S1. The comparison between these two groups identified the significant differences in tumor size, tumor location, Borrmann type, pN stage, pT stage, the number of positive lymph nodes and the ratio of positive. The results of univariate analysis identified that tumor size, tumor location, Borrmann type, Lauren type, lymphovascular invasion and pN stage were significant risk factors related to pT4a. Further multivariate analysis confirmed that tumor size, tumor location, Borrmann type, Lauren type, lymphovascular invasion and pN stage were significantly related to pT4a (Supplementary Table S2). Based on the results of multivariate analysis, we constructed a user-friendly RSS (Table 1). The score was assigned to each risk factor using the b-coefficients, as described in method. The median risk score was 5 (range 0e11). Patients were divided into 3 risk groups: low-risk (0e3 scores, 20.3%), medium-risk (4e6 scores,
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7 Table 1 Risk-scoring system to predict the probability of being diagnosed as pT4a for macroscopic serosal invasion positive patients. Risk factors
Tumor size Tumor location Borrmann type Lauren type Lymphovascular invasion UICC/AJCC N stage
Risk Scores 0
1
4 cm Lower/Middle Type I Intestinal Negative N0-1
>4 cm Upper
2 Entire Type II- III Diffuse Positive
3
Type IV
N2-3
53.0%), and high-risk group (7e11 scores, 26.7%). As the total scores increased, the actual rates and predicted probabilities of pT4a increased steadily (Fig. 1). Model evaluation and validation ROC analysis showed moderate-to-good discrimination ability of our model. The c-statistic estimated by using the bootstrap method was 0.698 (95% Confidence Interval (CI): 0.665e0.731) (Supplementary Fig. 2a). The Hosmer-Lemeshow test indicated the good fit of our model (P ¼ 0.521). The calibration curve showed predicted probabilities had good agreement with the actual rates (Supplementary Fig. 2b). The DCA showed that our RSS gained more net benefit than that of either treating all patients or none (Supplementary Fig. 2c). Therefore, MSI-positive patients using RSS to predict the depth of tumor invasion had superior net benefit than that of treating all MSI-positive patients as positive (T4a) or as negative (T3). Survival analysis and peritoneal recurrence assessment We compared the OS of MSI-positive pT3 patients in different risk groups with MSI-positive pT4a patients. (Fig. 2) The survival curves showed that pT3 patients in low- and medium-risk group exhibited better prognosis than pT3 patients in high-risk group and pT4a patients. However, pT3 patients in high-risk group had similar OS with pT4a patients (P ¼ 0.732). Because pN stage was included in this RSS and this might induce patients with higher pN stage to be assigned into the high-risk group and influence the prognosis
3
evaluation. Hence, we further compared the 5-year survival rates between pT3 and pT4a patients in each pN stage according to the different risk groups. Meanwhile, we also compared the 5-year survival rates between patients in different risk groups in each pN stage according to the pT stage. We combined low- and mediumrisk group together for analyses because they all had significant better prognosis than high-risk group and pT4a patients. Significant survival differences were observed between low-medium-risk group and high-risk group in pT3N1-3 subgroups, while no significant difference were observed in pT3N0 and pT4aN0-3 subgroups (Supplementary Table S3). These results indicated that MSIpositive pT3 patients in different risk groups should be treated differently, while pT4a patients did not. Moreover, in the lowmedium-risk group, the significant survival differences between pT3 and pT4a patients were observed in pN0-2, while not in pN3 subgroup. But no significant survival differences were observed between pT3 and pT4a patients in high-risk group regardless of pN stage. In addition, pT3 patients in high-risk group had similar prognosis with pT4a patients and worse prognosis than pT3 patients in low-medium-risk group in each pN stage (Supplementary Fig. 2aed). These results elucidated that pT3 patients in high-risk group had similar prognosis and should be treated as T4a. Based on above results, a novel revised-T stage was proposed, in which pT3 patients in high-risk group were incorporated into the T4a group. To investigate the prognostic significance of the revisedT stage and compare with the UICC/AJCC pT stage, two-step multivariate analyses were performed. Univariate analysis elucidated that tumor size, tumor location, Borrmann type, Lauren type, lymphovascular invasion, pN stage, MSI status, UICC/AJCC pT stage, revised-T stage and chemotherapy were significantly associated with OS (Supplementary Table S4). All the factors, except the revised-T stage, were included in the first step of multivariate analysis, the results showed that Borrmann type, Lauren type, lymphovascular invasion, UICC/AJCC pN stage, MSI status, UICC/ AJCC pT stage were independent prognostic factors. In the second step of multivariate analysis, the revised-T stage was included into the analysis. The results showed that the revised-T stage was one of the independent prognostic factors, whereas the UICC/AJCC pT stage and the MSI status were not (Table 2). The similar two-step multivariate analyses were also performed to evaluate the prediction significance of revised-T stage in peritoneal recurrence. In the univariate analysis, tumor size, Borrmann
Fig. 1. Patients were ranked by their total scores and then divided into 3 risk groups. The actual rates and predicted probabilities of pT4a in each score were presented. Scores 0 and 1, 10 and 11 were presented together due to small sample sizes.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
4
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
Fig. 2. Comparison overall survival curves between MSI-positive pT3 patients in different risk groups with MSI-positive pT4a patients. The P values represent the survival difference between each risk group of pT3 patients and pT4a patients. Table 2 Two-step multivariate analyses of prognostic factors for overall survival and the prediction of peritoneal recurrence. Overall survival
Peritoneal recurrence
HR
95%CI
P
RR
95%CI
P
1.295 1.271 1.181 1.371 1.183 1.354
1.143e1.467 1.114e1.450 1.031e1.353 1.294e1.453 1.004e1.393 1.181e1.552
<0.001 <0.001 0.016 <0.001 0.045 <0.001
1.378 1.978 e 1.553 1.673 2.226
1.017e1.865 1.444e2.709 e 1.358e1.775 1.124e2.490 1.607e3.084
0.038 <0.001 e <0.001 0.011 <0.001
1.273 1.208 1.380 1.504
1.124e1.441 1.055e1.382 1.304e1.461 1.312e1.724
<0.001 0.006 <0.001 <0.001
1.368 1.655 1.518 2.671
1.010e1.853 1.200e2.283 1.326e1.738 1.937e3.682
0.043 0.002 <0.001 <0.001
a
Step 1 Borrmann type Lauren type Lymphovascular invasion UICC/AJCC N stage MSI status UICC/AJCC T stage Step 2b Borrmann type Lauren type UICC/AJCC N stage Novel revised-T stage
HR: hazard ratio, CI: confidence interval, RR: relative risk, MSI: macroscopic serosal invasion. a Step 1All significant factors in univariate analysis were used for anaysis, excluding the novel revised-T stage. b Step2All significant factors in univariate analysis were used for anaysis, including the novel revised-T stage.
type, Lauren type, lymphovascular invasion, UICC/AJCC pN stage, MSI status, UICC/AJCC pT stage, revised-T stage and chemotherapy were significantly associated with peritoneal recurrence (Supplementary Table S4). All the significant predictors, except the revised-T stage, were included into the first step of multivariate analysis and Lauren type, UICC/AJCC pN stage, MSI status UICC/AJCC pT stage were identified as the independent predictors of peritoneal recurrence. When the revised-T stage incorporated into the second step of multivariate analysis, the revised-T stage was considered as one of the independent predictors while the UICC/ AJCC pT stage and MSI status were excluded from the prediction model. (Table 2).
Discussion Pathology examination is the gold standard for prognosis assessment. Surgical stage, which is evaluated by clinical surgeons intraoperatively, has also been suggested to be correlated with the prognosis [25]. However, the discordance of serosal invasion between pathological diagnosis and surgical evaluation often made surgeons difficult to have appropriate judgment for clinical therapy. In this study, we clarified the risk factors that were associated with pT4a for MSI-positive patients. Based on these factors, we established a RSS to assign patients into different risk groups and found that pT3 MSI-positive patient in high-risk group should be treated as T4a.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
In this study, the total agreement rate of serosal invasion between surgical evaluation and pathological diagnosis was 45.9%. Serosal invasion was underestimated in 55 (12.9%) pT4a patients. This was similar with the results of previous studies [2,5,25]. However, overestimation of serosal invasion was observed in 654 (74.6%) pT3 patients. This result was higher than previous studies [2,3,5,25]. Many previous studies investigated this phenomenon and provided several possible explanations [26e28]. These explanations mainly include two parts: the overestimation of surgical stage and the underestimation of pathological stage. When tumor invades to the subserosal layer, the normal structures of serosal layer are disrupted and have similar changes of serosal invasion while without any tumor cells spread into the serosal layer. This phenomenon is the so-called “overestimation of surgical stage”. It may induced by tumor-associated inflammation [27]. Underestimation of pathological stage, which is caused by missed detection of pathological examination, is another possible explanation of this discordance. Soga et al. reported that the prognosis of pT3 cancer with a wide width was similar to that of pT4a cancer and this phenomenon demonstrated that latent tumor exposure to serosal surface may happen if tumor width is wide [29]. In addition, more meticulous pathological examination may increase the detection rate of serosal invasion [26]. Thus, there is the possibility that the detection of serosal penetration between two adjacent section planes may be missed. In our present study, pT3 MSI-positive patients in high-risk group had similar survival with pT4a patients. And these patients also had higher possibility of being diagnosed as pT4a using RSS
5
(Fig. 1). Thus, we hypothesized that part of pT3 MSI-positive patients in high-risk group had the possibility of being underestimated. Therefore, we incorporated these patients into T4a group and a novel revised-T stage was formed. In the second step of multivariate analysis, the UICC/AJCC T stage was replaced by the novel revised-T stage. Interestingly, the MSI status was also disappeared in this step. These results elucidated that our novel revised-T stage, which combined pathological T stage with surgical T stage using this RSS, was more suitable for prognosis assessment than using these two stages independently. Based on the similar consideration with prognosis assessment, we also assessed the prediction ability of novel revised-T stage in peritoneal recurrence. The results of two-step multivariate analyses also suggested that the novel revised-T stage showed better prediction ability of peritoneal recurrence than pathological T stage and MSI status. The prediction implications of MSI status on prognosis and peritoneal recurrence have been reported in several studies [3,5,25,30]. Consistent with these studies, MSI status in our study was also the independent predictors of peritoneal recurrence and risk factors of prognosis (Table 2). Many studies also proposed some different MSI types for clinical implication [3,30]. Although these different types had good prediction ability, they were all proposed by single institution during their long time clinical experience and may not be widely used. Unlike these complex types, our RSS use much easier and more convenient method and can be broadly used and validated in other institutions. Although serial section takes long time and is complicated, it is still the most meticulous method for the detection of tumor depth.
Fig. 3. Proposed treatment algorithm for the tumor invasion reevaluation of MSI-positive patients.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
6
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
However, it is impossible to be applied in each patient, especially for high-volume cancer center. Thus, selecting appropriate patients receive this procedure could increase the detection rate of serosal invasion. In this study, patients in high-risk group had higher possibilities of being diagnosed as pT4a and the underestimation of serosal invasion may occur in these patients. Thus, serial section could perform for these patients to avoid the underestimation of pT stage and the stage migration. Based on above results, we proposed an algorithm for the treatment of MSI-positive patients (Fig. 3). Of the tumor invasion pathology-diagnosed pT3 that is evaluated as MSI-positive during the operation, we reevaluate the tumor invasion using our RSS. For individual patient, risk score is calculated after pathological assessment. For patient classified into low- and mediumrisk group, pathological evaluation remains the final diagnosis for treatment decision. However, patient in high-risk group should be treated as T4a and more invasive treatment and more meticulous pathological examination should be considered. Thus, our RSS provides a novel method to solve the discordance of serosal invasion between surgical evaluation and pathological diagnosis. However, there are numerous limitations in our study. Firstly, our RSS was established based on clinicopathological data from single institution, the conclusion might be biased by institutional practice patterns, especially the method of pathological examination. Secondly, we only performed internal validation to test our model and whether our model can be applied to other patient remained to be evaluated. External validation from other institutions should be performed in the future studies. Thirdly, all the prediction factors included in our RSS were availability only after surgical resection. This may affect the clinical applicability of our model. Including factors of preoperative examination such as gastroscopy, computed tomography and serum tumor markers (i.e.: CEA) could improve the clinical usefulness of our model. Thus, a prospective study cohort contained these preoperative predictors should be considered in our future studies. Conclusion In this study, we developed a RSS to predict the probability of tumor invasion diagnosed as pT4a for MSI-positive patients. Based on our RSS, MSI-positive pT3 patients in high-risk group should be treated as T4a when we make treatment decision and proposed treatment algorithm may be useful in clinical practice. Future studies should include other preoperative predictors to improve the clinical utility of our model. Conflict of interest statement All authors have no conflict of interests to disclose. Acknowledgments This work was supported by the National Natural Science Foundation of China (No. 81001092, No 81372550) and the Natural Science Foundation of Liaoning Province of China (No. 2013021097). Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.ejso.2018.01.240.
References [1] Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A. In: AJCC cancer staging manual. 7th ed. New York: Springer-Verlag; 2010. [2] Jeong O, Ryu SY, Jeong MR, Sun JW, Park YK. Accuracy of macroscopic intraoperative diagnosis of serosal invasion and risk of over- and underestimation in gastric carcinoma. World J Surg 2011;35(10):2252e8. [3] Kim JM, Jung H, Lee JS, Lee HH, Song KY, Park CH, et al. Clinical implication of serosal change in pathologic subserosa-limited gastric cancer. World J Surg 2012;36(2):355e61. [4] Korenaga D, Okuyama T, Orita H, Anai H, Baba H, Maehara Y, et al. Role of intraoperative assessment of lymph node metastasis and serosal invasion in patients with gastric cancer. J Surg Oncol 1994;55(4):250e4. [5] Yoo C, Ryu MH, Park YS, Yoo MW, Park SR, Ryoo BY, et al. Intraoperatively assessed macroscopic serosal changes in patients with curatively resected advanced gastric cancer: clinical implications for prognosis and peritoneal recurrence. Ann Surg Oncol 2015;22(9):2940e7. [6] Kang SY, Park HS, Kim CY. Prognostic significance of intraoperative macroscopic serosal invasion finding when it shows a discrepancy in pathologic result gastric cancer. Ann Surg Treat Res 2016;90(5):250e6. [7] Kim DJ, Lee JH, Kim W. Impact of intraoperative macroscopic diagnosis of serosal invasion in pathological subserosal (pT3) gastric cancer. J Gastric Cancer 2014;14(4):252e8. [8] Bando E, Kawamura T, Kinoshita K, Takahashi S, Maeda A, Osada S, et al. Magnitude of serosal changes predicts peritoneal recurrence of gastric cancer. J Am Coll Surg 2003;197(2):212e22. [9] Yoo CH, Noh SH, Shin DW, Choi SH, Min JS. Recurrence following curative resection for gastric carcinoma. Br J Surg 2000;87(2):236e42. [10] D'Angelica M, Gonen M, Brennan MF, Turnbull AD, Bains M, Karpeh MS. Patterns of initial recurrence in completely resected gastric adenocarcinoma. Ann Surg 2004;240(5):808e16. [11] Roviello F, Marrelli D, de Manzoni G, Morgagni P, Di Leo A, Saragoni L, et al. Prospective study of peritoneal recurrence after curative surgery for gastric cancer. Br J Surg 2003;90(9):1113e9. [12] Bando E, Yonemura Y, Takeshita Y, Taniguchi K, Yasui T, Yoshimitsu Y, et al. Intraoperative lavage for cytological examination in 1,297 patients with gastric carcinoma. Am J Surg 1999;178(3):256e62. [13] Kodera Y, Yamamura Y, Shimizu Y, Torii A, Hirai T, Yasui K, et al. Peritoneal washing cytology: prognostic value of positive findings in patients with gastric carcinoma undergoing a potentially curative resection. J Surg Oncol 1999;72(2):60e4. discussion 4e5. [14] Majima T, Ichikura T, Mochizuki H. Prognostic significance of the cytologic features of free cancer cells in the peritoneal cavity of patients with gastric cancer. Surg Today 2002;32(1):35e9. [15] Jonas S, Weinrich M, Tullius SG, Al-Abadi H, Steinbrich R, Radke C, et al. Microscopic tumor cell dissemination in gastric cancer. Surg Today 2004; 34(2):101e6. [16] Wang JY, Lin SR, Lu CY, Chen CC, Wu DC, Chai CY, et al. Gastric cancer cell detection in peritoneal lavage: RT-PCR for carcinoembryonic antigen transcripts versus the combined cytology with peritoneal carcinoembryonic antigen levels. Cancer Lett 2005;223(1):129e35. [17] Sun J, Song Y, Wang Z, Gao P, Chen X, Xu Y, et al. Benefits of hyperthermic intraperitoneal chemotherapy for patients with serosal invasion in gastric cancer: a meta-analysis of the randomized controlled trials. BMC Cancer 2012;12:526. [18] Kang YK, Yook JH, Chang HM, Ryu MH, Yoo C, Zang DY, et al. Enhanced efficacy of postoperative adjuvant chemotherapy in advanced gastric cancer: results from a phase 3 randomized trial (AMC0101). Cancer Chemother Pharmacol 2014;73(1):139e49. [19] Miyashiro I, Furukawa H, Sasako M, Yamamoto S, Nashimoto A, Nakajima T, et al. Randomized clinical trial of adjuvant chemotherapy with intraperitoneal and intravenous cisplatin followed by oral fluorouracil (UFT) in serosapositive gastric cancer versus curative resection alone: final results of the Japan Clinical Oncology Group trial JCOG9206-2. Gastric Cancer 2011;14(3): 212e8. [20] Yan TD, Black D, Sugarbaker PH, Zhu J, Yonemura Y, Petrou G, et al. A systematic review and meta-analysis of the randomized controlled trials on adjuvant intraperitoneal chemotherapy for resectable gastric cancer. Ann Surg Oncol 2007;14(10):2702e13. [21] Ichiyoshi Y, Maehara Y, Tomisaki S, Oiwa H, Sakaguchi Y, Ohno S, et al. Macroscopic intraoperative diagnosis of serosal invasion and clinical outcome of gastric cancer: risk of underestimation. J Surg Oncol 1995;59(4):255e60. [22] Kaminski MF, Polkowski M, Kraszewska E, Rupinski M, Butruk E, Regula J. A score to estimate the likelihood of detecting advanced colorectal neoplasia at colonoscopy. Gut 2014;63(7):1112e9. [23] Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143(1):29e36. [24] Vickers AJ, Cronin AM, Elkin EB, Gonen M. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers. BMC Med Inf Decis Making 2008;8:53. [25] Lee SE, Ryu KW, Nam BH, Lee JH, Choi IJ, Kook MC, et al. Prognostic significance of intraoperatively estimated surgical stage in curatively resected gastric cancer patients. J Am Coll Surg 2009;209(4):461e7.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7 [26] Stewart CJ, Hillery S, Platell C, Puppa G. Assessment of serosal invasion and criteria for the classification of pathological (p) T4 staging in colorectal carcinoma: confusions, controversies and criticisms. Cancers (Basel) 2011;3(1): 164e81. [27] Stewart C, Hillery S, Havlat M. Serosal involvement in colorectal carcinoma. Histopathology 2006;49(4):435e7. [28] Ludeman L, Shepherd NA. Serosal involvement in gastrointestinal cancer: its assessment and significance. Histopathology 2005;47(2):123e31.
7
[29] Soga K, Ichikawa D, Yasukawa S, Kubota T, Kikuchi S, Fujiwara H, et al. Prognostic impact of the width of subserosal invasion in gastric cancer invading the subserosal layer. Surgery 2010;147(2):197e203. [30] Sun Z, Xu YY, Wang ZN, Zhu Z, Zhang H, Huang BJ, et al. Macroscopic serosal classification predicts peritoneal recurrence for patients with gastric cancer underwent potentially curative surgery. Ann Surg Oncol 2011;18(4): 1068e80.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
Contents lists available at ScienceDirect
European Journal of Surgical Oncology journal homepage: www.ejso.com
Consensus Statement
Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients Peng-liang Wang, Jin-yu Huang, Zhi Zhu, Bao-cheng Gong, Han-wei Huang, Shi-jie Duan, Hui-mian Xu**, Fu-nan Liu* Department of Surgical Oncology, The First Affiliated Hospital of China Medical University, Shenyang, China
a r t i c l e i n f o
a b s t r a c t
Article history: Accepted 30 January 2018 Available online xxx
Background: The status of serosal invasion is often discordance between pathological and intraoperative evaluation. Our study sought to develop a risk-scoring system (RSS) to predict the probability of pT4a for macroscopic serosal invasion (MSI) positive patients and reevaluate the serosal invasion status. Patients and Methods: A total of 1301 pT3/pT4a gastric cancer patients with curative surgery were reviewed. We constructed the RSS to predict the probability of pT4a and assigned MSI-positive patients into different risk groups based on the risk scores. The prognostic significance of these risk groups was also evaluated. Results: Univariate and multivariate analyses identified that tumor location, Lauren type, Borrmann type, tumor size, lymphovascular invasion and pN stage were risk factors related to pT4a. Survival analyses showed that pT3 MSI-positive patients in high-risk group had similar survival with pT4a patients. We incorporated these two groups into one stage and proposed a novel revised-T stage. Two-step multivariate analyses indicated that the revised-T stage showed better prediction ability for prognosis and peritoneal recurrence assessment than original pT stage and MSI status. Conclusions: In our present study, we developed a RSS to predict the probability of pT4a for MSI-positive patients. Based on our RSS, we proposed a treatment algorithm to reevaluate the tumor invasion for MSIpositive patients in clinical practice. Future studies should include other preoperative predictors to improve the clinical utility of our model. © 2018 Elsevier Ltd, BASO ~ The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved.
Keywords: Gastric cancer Macroscopic serosal invasion Risk-scoring system Peritoneal recurrence
Introduction Recently, American Joint Committee on Cancer/International Union Against Cancer (AJCC/UICC) tumor-node-metastasis (TNM) classification, which based on the depth of tumor invasion, number of regional metastasis lymph nodes and distant metastasis, is the most widely used staging of gastric cancer (GC) [1]. However, the pathological T stage is often inconsistent with the T stage evaluated by surgeons during operations, especially the status of serosal
* Corresponding author. Department of Surgical Oncology, The First Affiliated Hospital of China Medical University, 155 Nanjing North Street, Heping District, Shenyang, 110001, China. ** Corresponding author. Department of Surgical Oncology, The First Affiliated Hospital of China Medical University, 155 Nanjing North Street, Heping District, Shenyang, 110001, China. E-mail addresses: [email protected] (H.-m. Xu), [email protected] (F.-n. Liu).
invasion [2e4]. And about half of the patients were macroscopic serosal invasion (MSI) positive and finally identified as the serosallimited tumor invasion pathologically [4e6]. Several studies also demonstrated that MSI-positive patients had worse prognosis as well as higher incidence of peritoneal recurrence [3,5e8]. Peritoneal recurrence is the most common pattern of recurrence and almost half of advanced GC patients died of peritoneal recurrence even after curative resection [9e11]. Previous studies indicated that intraperitoneal free cancer cells are significantly associated with peritoneal recurrence [12e15]. When tumor invades into the serosal layer, cancer cells may exfoliate from the lesion where the normal serosal are disrupted by tumor invasion [8,16]. Thus, the status of serosal invasion is extremely critical for detection of intraperitoneal free cancer cells and prediction of peritoneal recurrence. Although hyperthermic intraperitoneal chemotherapy (HIPC) may improve the overall survival (OS) and decrease the peritoneal
https://doi.org/10.1016/j.ejso.2018.01.240 0748-7983/© 2018 Elsevier Ltd, BASO ~ The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
2
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
recurrence among patients with pathological serosal invasion [17], the efficacy of this treatment for MSI-positive patients remains controversial [18,19]. Considering the morbidity and mortality rate of this treatment, the accurate assessment of tumor invasion is prerequisite for these patients [20]. Based on above considerations, we sought to reevaluate the tumor invasion status for these patients by developing a riskscoring system (RSS).
evaluated by plotting the calibration curve and testing the statistical significance of Hosmer-Lemeshow goodness-of-fit test (P > 0.05 indicating good fit). Both of the c-statistic and the calibration curve were internally validated by bootstrap method to reduce the over fit bias (200 repetitions). Decision curve analysis (DCA) was constructed to evaluate the clinical usefulness of RSS [24]. This analysis estimates the clinical net benefit of our model by summing the benefits (true-positive) and subtracting the harms (false-positive).
Patients and Methods Survival analysis and peritoneal recurrence assessment Study population A total of 2062 GC patients who underwent gastric resection and combined D2/D3 lymphadenectomy and achieved R0 resection at the Department of Surgical Oncology, the First Affiliated Hospital of China Medical University from January 1987 to December 2006 were collected. Of these patients, 1366 patients were confirmed as pT3/pT4a pathologically according to the seventh edition of the UICC/AJCC TNM staging system [1]. We mainly included these patients for analyses because the majority of MSI-positive patients were finally identified as pT3/pT4a histologically [2,4,6,21]. Among 1366 patients, 28 patients died in the postoperative period (<3 months). All patients received follow-up after being discharged from hospital, every quarter of the first 3-year, 6 months during the fourth and fifth year, and once a year thereafter. Followup evaluations included patient history, physical examination, carcinoembryonic antigen (CEA), carbohydrate antigen19-9 (CA199) and CA72-4. The gastroscopy and computed tomography (CT) scan were performed once a year. At the time of the last follow-up (March 31, 2011), the median and mean follow-up time was 32.0 and 61.8 months, respectively (range 3e313 months). 24 patients were lost and these patients were excluded from our study (followup rate: 98.2%). Of the remaining 1314 patients, 13 patients with incomplete information were also excluded. Finally, 1301 patients were selected for further analysis. A total of 753 patients received postoperative intravenous adjuvant chemotherapy and/or intraperitoneal chemotherapy. No one received neoadjuvant chemotherapy and/or radiotherapy. At the time of the last follow-up, 622 patients died of cancer-related diseases and 362 patients died of other diseases or causes unknown. Peritoneal recurrence was diagnosed mainly by CT scan and sometimes identified by cytology of ascites or intraoperative biopsy if operations were performed. Of the 622 patients, 293 were identified with peritoneal recurrence. This study was approved by the Institutional Ethics Committee of China Medical University. Development and evaluation of risk-scoring system Logistic regression analysis was used to identify the risk factors of tumor invasion diagnosed as pT4a. The score of each risk factor was estimated based on the results of multivariate analysis. The bcoefficients of multivariate analysis were rounded up to the nearest half integer and then multiplied by 2 to avoid decimals [22]. We calculated the total points of each patient and divided the patients into quartiles approximately. Then three risk groups (low-, medium-, and high-risk group) were established with the middle two quarters combined. The total risk score of each patient was used as a risk factor for the model evaluation. The performance of the developed RSS was evaluated with respect to discrimination and calibration. Discrimination was measured by the concordance statistic (c-statistic), which is equivalent to the area under the receiver operating characteristic (ROC) curve [23]. The calibration of the model was
To elucidate the prognosis influence of our RSS, we compared the OS between these risk groups in each pN stage and pT stage. Meanwhile, we compared the OS of pT3 MSI-positive patients in different risk groups with pT4a MSI-positive patients. Based on these results, patients who had similar survival were incorporated into one T stage and a novel revised-T stage was formed. Additionally, two-step multivariate analysis was also used to compare the novel revised-T stage with original UICC/AJCC pT stage in the prediction of peritoneal recurrence and the prognosis assessment. Patients died of other diseases or with causes unknown (n ¼ 362) were excluded when we evaluated the peritoneal recurrence. Statistical analysis Continuous variates were described as mean ± standard deviation (SD) and categorical variates were described as count and percentages. Comparisons between groups were performed by using Pearson c2 test for categorical variates and two-tailed Student's t-test for continuous variates. Univariate analyses were performed to indentify the significant risk factors related with pT4a and peritoneal recurrence. Multivariate analyses were used to identify the independent risk factors. The OS rates were estimated by Kaplan-Meier curves and the comparisons were identified by the log-rank test. Multivariate Cox proportional hazards regression model was used to identify the independent prognosis predictors. Only those variates with P values < 0.05 in univariate analyses were selected into the multivariate analyses. The analyses were performed using the R software version 3.2.0 (http://www.r-project.org/) and SPSS version 21.0 (SPSS, Chicago, IL). A two-tailed P value < 0.05 was considered as statistically significant in all analyses. Results Study population and model development Clinicopathological characteristics of MSI-positive and -negative patients were presented in Supplementary Table S1. The comparison between these two groups identified the significant differences in tumor size, tumor location, Borrmann type, pN stage, pT stage, the number of positive lymph nodes and the ratio of positive. The results of univariate analysis identified that tumor size, tumor location, Borrmann type, Lauren type, lymphovascular invasion and pN stage were significant risk factors related to pT4a. Further multivariate analysis confirmed that tumor size, tumor location, Borrmann type, Lauren type, lymphovascular invasion and pN stage were significantly related to pT4a (Supplementary Table S2). Based on the results of multivariate analysis, we constructed a user-friendly RSS (Table 1). The score was assigned to each risk factor using the b-coefficients, as described in method. The median risk score was 5 (range 0e11). Patients were divided into 3 risk groups: low-risk (0e3 scores, 20.3%), medium-risk (4e6 scores,
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7 Table 1 Risk-scoring system to predict the probability of being diagnosed as pT4a for macroscopic serosal invasion positive patients. Risk factors
Tumor size Tumor location Borrmann type Lauren type Lymphovascular invasion UICC/AJCC N stage
Risk Scores 0
1
4 cm Lower/Middle Type I Intestinal Negative N0-1
>4 cm Upper
2 Entire Type II- III Diffuse Positive
3
Type IV
N2-3
53.0%), and high-risk group (7e11 scores, 26.7%). As the total scores increased, the actual rates and predicted probabilities of pT4a increased steadily (Fig. 1). Model evaluation and validation ROC analysis showed moderate-to-good discrimination ability of our model. The c-statistic estimated by using the bootstrap method was 0.698 (95% Confidence Interval (CI): 0.665e0.731) (Supplementary Fig. 2a). The Hosmer-Lemeshow test indicated the good fit of our model (P ¼ 0.521). The calibration curve showed predicted probabilities had good agreement with the actual rates (Supplementary Fig. 2b). The DCA showed that our RSS gained more net benefit than that of either treating all patients or none (Supplementary Fig. 2c). Therefore, MSI-positive patients using RSS to predict the depth of tumor invasion had superior net benefit than that of treating all MSI-positive patients as positive (T4a) or as negative (T3). Survival analysis and peritoneal recurrence assessment We compared the OS of MSI-positive pT3 patients in different risk groups with MSI-positive pT4a patients. (Fig. 2) The survival curves showed that pT3 patients in low- and medium-risk group exhibited better prognosis than pT3 patients in high-risk group and pT4a patients. However, pT3 patients in high-risk group had similar OS with pT4a patients (P ¼ 0.732). Because pN stage was included in this RSS and this might induce patients with higher pN stage to be assigned into the high-risk group and influence the prognosis
3
evaluation. Hence, we further compared the 5-year survival rates between pT3 and pT4a patients in each pN stage according to the different risk groups. Meanwhile, we also compared the 5-year survival rates between patients in different risk groups in each pN stage according to the pT stage. We combined low- and mediumrisk group together for analyses because they all had significant better prognosis than high-risk group and pT4a patients. Significant survival differences were observed between low-medium-risk group and high-risk group in pT3N1-3 subgroups, while no significant difference were observed in pT3N0 and pT4aN0-3 subgroups (Supplementary Table S3). These results indicated that MSIpositive pT3 patients in different risk groups should be treated differently, while pT4a patients did not. Moreover, in the lowmedium-risk group, the significant survival differences between pT3 and pT4a patients were observed in pN0-2, while not in pN3 subgroup. But no significant survival differences were observed between pT3 and pT4a patients in high-risk group regardless of pN stage. In addition, pT3 patients in high-risk group had similar prognosis with pT4a patients and worse prognosis than pT3 patients in low-medium-risk group in each pN stage (Supplementary Fig. 2aed). These results elucidated that pT3 patients in high-risk group had similar prognosis and should be treated as T4a. Based on above results, a novel revised-T stage was proposed, in which pT3 patients in high-risk group were incorporated into the T4a group. To investigate the prognostic significance of the revisedT stage and compare with the UICC/AJCC pT stage, two-step multivariate analyses were performed. Univariate analysis elucidated that tumor size, tumor location, Borrmann type, Lauren type, lymphovascular invasion, pN stage, MSI status, UICC/AJCC pT stage, revised-T stage and chemotherapy were significantly associated with OS (Supplementary Table S4). All the factors, except the revised-T stage, were included in the first step of multivariate analysis, the results showed that Borrmann type, Lauren type, lymphovascular invasion, UICC/AJCC pN stage, MSI status, UICC/ AJCC pT stage were independent prognostic factors. In the second step of multivariate analysis, the revised-T stage was included into the analysis. The results showed that the revised-T stage was one of the independent prognostic factors, whereas the UICC/AJCC pT stage and the MSI status were not (Table 2). The similar two-step multivariate analyses were also performed to evaluate the prediction significance of revised-T stage in peritoneal recurrence. In the univariate analysis, tumor size, Borrmann
Fig. 1. Patients were ranked by their total scores and then divided into 3 risk groups. The actual rates and predicted probabilities of pT4a in each score were presented. Scores 0 and 1, 10 and 11 were presented together due to small sample sizes.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
4
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
Fig. 2. Comparison overall survival curves between MSI-positive pT3 patients in different risk groups with MSI-positive pT4a patients. The P values represent the survival difference between each risk group of pT3 patients and pT4a patients. Table 2 Two-step multivariate analyses of prognostic factors for overall survival and the prediction of peritoneal recurrence. Overall survival
Peritoneal recurrence
HR
95%CI
P
RR
95%CI
P
1.295 1.271 1.181 1.371 1.183 1.354
1.143e1.467 1.114e1.450 1.031e1.353 1.294e1.453 1.004e1.393 1.181e1.552
<0.001 <0.001 0.016 <0.001 0.045 <0.001
1.378 1.978 e 1.553 1.673 2.226
1.017e1.865 1.444e2.709 e 1.358e1.775 1.124e2.490 1.607e3.084
0.038 <0.001 e <0.001 0.011 <0.001
1.273 1.208 1.380 1.504
1.124e1.441 1.055e1.382 1.304e1.461 1.312e1.724
<0.001 0.006 <0.001 <0.001
1.368 1.655 1.518 2.671
1.010e1.853 1.200e2.283 1.326e1.738 1.937e3.682
0.043 0.002 <0.001 <0.001
a
Step 1 Borrmann type Lauren type Lymphovascular invasion UICC/AJCC N stage MSI status UICC/AJCC T stage Step 2b Borrmann type Lauren type UICC/AJCC N stage Novel revised-T stage
HR: hazard ratio, CI: confidence interval, RR: relative risk, MSI: macroscopic serosal invasion. a Step 1All significant factors in univariate analysis were used for anaysis, excluding the novel revised-T stage. b Step2All significant factors in univariate analysis were used for anaysis, including the novel revised-T stage.
type, Lauren type, lymphovascular invasion, UICC/AJCC pN stage, MSI status, UICC/AJCC pT stage, revised-T stage and chemotherapy were significantly associated with peritoneal recurrence (Supplementary Table S4). All the significant predictors, except the revised-T stage, were included into the first step of multivariate analysis and Lauren type, UICC/AJCC pN stage, MSI status UICC/AJCC pT stage were identified as the independent predictors of peritoneal recurrence. When the revised-T stage incorporated into the second step of multivariate analysis, the revised-T stage was considered as one of the independent predictors while the UICC/ AJCC pT stage and MSI status were excluded from the prediction model. (Table 2).
Discussion Pathology examination is the gold standard for prognosis assessment. Surgical stage, which is evaluated by clinical surgeons intraoperatively, has also been suggested to be correlated with the prognosis [25]. However, the discordance of serosal invasion between pathological diagnosis and surgical evaluation often made surgeons difficult to have appropriate judgment for clinical therapy. In this study, we clarified the risk factors that were associated with pT4a for MSI-positive patients. Based on these factors, we established a RSS to assign patients into different risk groups and found that pT3 MSI-positive patient in high-risk group should be treated as T4a.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
In this study, the total agreement rate of serosal invasion between surgical evaluation and pathological diagnosis was 45.9%. Serosal invasion was underestimated in 55 (12.9%) pT4a patients. This was similar with the results of previous studies [2,5,25]. However, overestimation of serosal invasion was observed in 654 (74.6%) pT3 patients. This result was higher than previous studies [2,3,5,25]. Many previous studies investigated this phenomenon and provided several possible explanations [26e28]. These explanations mainly include two parts: the overestimation of surgical stage and the underestimation of pathological stage. When tumor invades to the subserosal layer, the normal structures of serosal layer are disrupted and have similar changes of serosal invasion while without any tumor cells spread into the serosal layer. This phenomenon is the so-called “overestimation of surgical stage”. It may induced by tumor-associated inflammation [27]. Underestimation of pathological stage, which is caused by missed detection of pathological examination, is another possible explanation of this discordance. Soga et al. reported that the prognosis of pT3 cancer with a wide width was similar to that of pT4a cancer and this phenomenon demonstrated that latent tumor exposure to serosal surface may happen if tumor width is wide [29]. In addition, more meticulous pathological examination may increase the detection rate of serosal invasion [26]. Thus, there is the possibility that the detection of serosal penetration between two adjacent section planes may be missed. In our present study, pT3 MSI-positive patients in high-risk group had similar survival with pT4a patients. And these patients also had higher possibility of being diagnosed as pT4a using RSS
5
(Fig. 1). Thus, we hypothesized that part of pT3 MSI-positive patients in high-risk group had the possibility of being underestimated. Therefore, we incorporated these patients into T4a group and a novel revised-T stage was formed. In the second step of multivariate analysis, the UICC/AJCC T stage was replaced by the novel revised-T stage. Interestingly, the MSI status was also disappeared in this step. These results elucidated that our novel revised-T stage, which combined pathological T stage with surgical T stage using this RSS, was more suitable for prognosis assessment than using these two stages independently. Based on the similar consideration with prognosis assessment, we also assessed the prediction ability of novel revised-T stage in peritoneal recurrence. The results of two-step multivariate analyses also suggested that the novel revised-T stage showed better prediction ability of peritoneal recurrence than pathological T stage and MSI status. The prediction implications of MSI status on prognosis and peritoneal recurrence have been reported in several studies [3,5,25,30]. Consistent with these studies, MSI status in our study was also the independent predictors of peritoneal recurrence and risk factors of prognosis (Table 2). Many studies also proposed some different MSI types for clinical implication [3,30]. Although these different types had good prediction ability, they were all proposed by single institution during their long time clinical experience and may not be widely used. Unlike these complex types, our RSS use much easier and more convenient method and can be broadly used and validated in other institutions. Although serial section takes long time and is complicated, it is still the most meticulous method for the detection of tumor depth.
Fig. 3. Proposed treatment algorithm for the tumor invasion reevaluation of MSI-positive patients.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
6
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7
However, it is impossible to be applied in each patient, especially for high-volume cancer center. Thus, selecting appropriate patients receive this procedure could increase the detection rate of serosal invasion. In this study, patients in high-risk group had higher possibilities of being diagnosed as pT4a and the underestimation of serosal invasion may occur in these patients. Thus, serial section could perform for these patients to avoid the underestimation of pT stage and the stage migration. Based on above results, we proposed an algorithm for the treatment of MSI-positive patients (Fig. 3). Of the tumor invasion pathology-diagnosed pT3 that is evaluated as MSI-positive during the operation, we reevaluate the tumor invasion using our RSS. For individual patient, risk score is calculated after pathological assessment. For patient classified into low- and mediumrisk group, pathological evaluation remains the final diagnosis for treatment decision. However, patient in high-risk group should be treated as T4a and more invasive treatment and more meticulous pathological examination should be considered. Thus, our RSS provides a novel method to solve the discordance of serosal invasion between surgical evaluation and pathological diagnosis. However, there are numerous limitations in our study. Firstly, our RSS was established based on clinicopathological data from single institution, the conclusion might be biased by institutional practice patterns, especially the method of pathological examination. Secondly, we only performed internal validation to test our model and whether our model can be applied to other patient remained to be evaluated. External validation from other institutions should be performed in the future studies. Thirdly, all the prediction factors included in our RSS were availability only after surgical resection. This may affect the clinical applicability of our model. Including factors of preoperative examination such as gastroscopy, computed tomography and serum tumor markers (i.e.: CEA) could improve the clinical usefulness of our model. Thus, a prospective study cohort contained these preoperative predictors should be considered in our future studies. Conclusion In this study, we developed a RSS to predict the probability of tumor invasion diagnosed as pT4a for MSI-positive patients. Based on our RSS, MSI-positive pT3 patients in high-risk group should be treated as T4a when we make treatment decision and proposed treatment algorithm may be useful in clinical practice. Future studies should include other preoperative predictors to improve the clinical utility of our model. Conflict of interest statement All authors have no conflict of interests to disclose. Acknowledgments This work was supported by the National Natural Science Foundation of China (No. 81001092, No 81372550) and the Natural Science Foundation of Liaoning Province of China (No. 2013021097). Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.ejso.2018.01.240.
References [1] Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A. In: AJCC cancer staging manual. 7th ed. New York: Springer-Verlag; 2010. [2] Jeong O, Ryu SY, Jeong MR, Sun JW, Park YK. Accuracy of macroscopic intraoperative diagnosis of serosal invasion and risk of over- and underestimation in gastric carcinoma. World J Surg 2011;35(10):2252e8. [3] Kim JM, Jung H, Lee JS, Lee HH, Song KY, Park CH, et al. Clinical implication of serosal change in pathologic subserosa-limited gastric cancer. World J Surg 2012;36(2):355e61. [4] Korenaga D, Okuyama T, Orita H, Anai H, Baba H, Maehara Y, et al. Role of intraoperative assessment of lymph node metastasis and serosal invasion in patients with gastric cancer. J Surg Oncol 1994;55(4):250e4. [5] Yoo C, Ryu MH, Park YS, Yoo MW, Park SR, Ryoo BY, et al. Intraoperatively assessed macroscopic serosal changes in patients with curatively resected advanced gastric cancer: clinical implications for prognosis and peritoneal recurrence. Ann Surg Oncol 2015;22(9):2940e7. [6] Kang SY, Park HS, Kim CY. Prognostic significance of intraoperative macroscopic serosal invasion finding when it shows a discrepancy in pathologic result gastric cancer. Ann Surg Treat Res 2016;90(5):250e6. [7] Kim DJ, Lee JH, Kim W. Impact of intraoperative macroscopic diagnosis of serosal invasion in pathological subserosal (pT3) gastric cancer. J Gastric Cancer 2014;14(4):252e8. [8] Bando E, Kawamura T, Kinoshita K, Takahashi S, Maeda A, Osada S, et al. Magnitude of serosal changes predicts peritoneal recurrence of gastric cancer. J Am Coll Surg 2003;197(2):212e22. [9] Yoo CH, Noh SH, Shin DW, Choi SH, Min JS. Recurrence following curative resection for gastric carcinoma. Br J Surg 2000;87(2):236e42. [10] D'Angelica M, Gonen M, Brennan MF, Turnbull AD, Bains M, Karpeh MS. Patterns of initial recurrence in completely resected gastric adenocarcinoma. Ann Surg 2004;240(5):808e16. [11] Roviello F, Marrelli D, de Manzoni G, Morgagni P, Di Leo A, Saragoni L, et al. Prospective study of peritoneal recurrence after curative surgery for gastric cancer. Br J Surg 2003;90(9):1113e9. [12] Bando E, Yonemura Y, Takeshita Y, Taniguchi K, Yasui T, Yoshimitsu Y, et al. Intraoperative lavage for cytological examination in 1,297 patients with gastric carcinoma. Am J Surg 1999;178(3):256e62. [13] Kodera Y, Yamamura Y, Shimizu Y, Torii A, Hirai T, Yasui K, et al. Peritoneal washing cytology: prognostic value of positive findings in patients with gastric carcinoma undergoing a potentially curative resection. J Surg Oncol 1999;72(2):60e4. discussion 4e5. [14] Majima T, Ichikura T, Mochizuki H. Prognostic significance of the cytologic features of free cancer cells in the peritoneal cavity of patients with gastric cancer. Surg Today 2002;32(1):35e9. [15] Jonas S, Weinrich M, Tullius SG, Al-Abadi H, Steinbrich R, Radke C, et al. Microscopic tumor cell dissemination in gastric cancer. Surg Today 2004; 34(2):101e6. [16] Wang JY, Lin SR, Lu CY, Chen CC, Wu DC, Chai CY, et al. Gastric cancer cell detection in peritoneal lavage: RT-PCR for carcinoembryonic antigen transcripts versus the combined cytology with peritoneal carcinoembryonic antigen levels. Cancer Lett 2005;223(1):129e35. [17] Sun J, Song Y, Wang Z, Gao P, Chen X, Xu Y, et al. Benefits of hyperthermic intraperitoneal chemotherapy for patients with serosal invasion in gastric cancer: a meta-analysis of the randomized controlled trials. BMC Cancer 2012;12:526. [18] Kang YK, Yook JH, Chang HM, Ryu MH, Yoo C, Zang DY, et al. Enhanced efficacy of postoperative adjuvant chemotherapy in advanced gastric cancer: results from a phase 3 randomized trial (AMC0101). Cancer Chemother Pharmacol 2014;73(1):139e49. [19] Miyashiro I, Furukawa H, Sasako M, Yamamoto S, Nashimoto A, Nakajima T, et al. Randomized clinical trial of adjuvant chemotherapy with intraperitoneal and intravenous cisplatin followed by oral fluorouracil (UFT) in serosapositive gastric cancer versus curative resection alone: final results of the Japan Clinical Oncology Group trial JCOG9206-2. Gastric Cancer 2011;14(3): 212e8. [20] Yan TD, Black D, Sugarbaker PH, Zhu J, Yonemura Y, Petrou G, et al. A systematic review and meta-analysis of the randomized controlled trials on adjuvant intraperitoneal chemotherapy for resectable gastric cancer. Ann Surg Oncol 2007;14(10):2702e13. [21] Ichiyoshi Y, Maehara Y, Tomisaki S, Oiwa H, Sakaguchi Y, Ohno S, et al. Macroscopic intraoperative diagnosis of serosal invasion and clinical outcome of gastric cancer: risk of underestimation. J Surg Oncol 1995;59(4):255e60. [22] Kaminski MF, Polkowski M, Kraszewska E, Rupinski M, Butruk E, Regula J. A score to estimate the likelihood of detecting advanced colorectal neoplasia at colonoscopy. Gut 2014;63(7):1112e9. [23] Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143(1):29e36. [24] Vickers AJ, Cronin AM, Elkin EB, Gonen M. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers. BMC Med Inf Decis Making 2008;8:53. [25] Lee SE, Ryu KW, Nam BH, Lee JH, Choi IJ, Kook MC, et al. Prognostic significance of intraoperatively estimated surgical stage in curatively resected gastric cancer patients. J Am Coll Surg 2009;209(4):461e7.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240
P.-l. Wang et al. / European Journal of Surgical Oncology xxx (2018) 1e7 [26] Stewart CJ, Hillery S, Platell C, Puppa G. Assessment of serosal invasion and criteria for the classification of pathological (p) T4 staging in colorectal carcinoma: confusions, controversies and criticisms. Cancers (Basel) 2011;3(1): 164e81. [27] Stewart C, Hillery S, Havlat M. Serosal involvement in colorectal carcinoma. Histopathology 2006;49(4):435e7. [28] Ludeman L, Shepherd NA. Serosal involvement in gastrointestinal cancer: its assessment and significance. Histopathology 2005;47(2):123e31.
7
[29] Soga K, Ichikawa D, Yasukawa S, Kubota T, Kikuchi S, Fujiwara H, et al. Prognostic impact of the width of subserosal invasion in gastric cancer invading the subserosal layer. Surgery 2010;147(2):197e203. [30] Sun Z, Xu YY, Wang ZN, Zhu Z, Zhang H, Huang BJ, et al. Macroscopic serosal classification predicts peritoneal recurrence for patients with gastric cancer underwent potentially curative surgery. Ann Surg Oncol 2011;18(4): 1068e80.
Please cite this article in press as: Wang P-l, et al., Development of a risk-scoring system to evaluate the serosal invasion for macroscopic serosal invasion positive gastric cancer patients, European Journal of Surgical Oncology (2018), https://doi.org/10.1016/j.ejso.2018.01.240