Abstract
Background/Aim: Peritoneal cancer index (PCI) has been a strong indicator of prognosis for patients receiving cytoreductive surgery. The aim of this single institution study was to compare the survival of peritoneal carcinoma cases treated with cytoreduction surgery arising from colorectal cancer grouped by PCI scores. Patients and Methods: A retrospective study of a prospective dataset maintained from 2000 till September 2022, for peritoneal metastases from colorectal cancers was carried out. Of the total of 1,625 peritoneal metastases cases, 415 were identified with colorectal cancer and considered for analysis. Survival was followed for 60 months since the index-peritonectomy for cases in this study. Results: Hazard ratio for 5-year survival using the Cox regression analysis over time (t) with a Log-rank (Mantel–Cox) test for significance between the groups indicated, <15 vs. 15 (HR=2.121, p=0.0338), <15 vs. 16-20 (HR=2.748, p<0.0001), <15 vs. >20 (HR=3.158, p<0.0001), 15 vs. 16-20 (HR=1.262, p=0.5658), 15 vs. >20 (HR=1.566, p=0.2771) and for PCI category 16-20 vs. >20 (HR=1.204, p=0.5355) for survival. Median survival for the categories of PCI <15, PCI-15, PCI-16-20, and PCI >15 was 43.967 (95%CI=28.31-59.63), 20.67 (95%CI=5.01-36.33), 19.50 (95%CI=3.84-35.16), and 14.30 (95%CI=1.36-29.96), respectively. Conclusion: A correlation of PCI with survival was confirmed in this study reinforcing the need for assessment of PCI at surgery to help prognostication. Detecting synchronous peritoneal metastases early and prompt treatment can help prevent recurrence and increase survival.
Long term survival of patients with colorectal cancer depends on two main factors, the aggressiveness of the cancer and its visceral spread. Dissemination of the cancer via lymph nodes is common resulting in peritoneal spread of the cancer and liver metastases, which have a central role in overall prognosis (1). Recently, early diagnosis of peritoneal metastases has helped in identifying surgical treatment options and it is known that complete cytoreduction at surgery with nil macroscopic residual disease (cc0) is a prerequisite for survival (1). The 5-year survival of patients with colorectal cancers with limited peritoneal metastases receiving cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) has improved, ranging in many studies from 11% to 51% (2, 3). Historically systemic treatments have had little effect on peritoneal carcinomatosis (4) and although treatment with oxaliplatin and anti-angiogenesis agents has improved survival, the median overall survival (OS) has not exceeded more than 18 months (5). However, extensive cytoreduction with complete removal of these metastases for suitable cases with a completeness of cytoreduction score (CC-score) of 0 is necessary to achieve a favourable survival outcome (3, 4).
Peritoneal cancer index (PCI) has been a strong indicator of prognosis for patients receiving cytoreductive surgery. However, radiological determination of PCI before surgery has often not correlated with the surgical PCI due to the limitations in visualisation of the tumour (4, 6-8). In addition, several clinicopathological variables impact the prognosis of colorectal carcinoma. Lymph node involvement, tumour differentiation, presence of mucin and CC-score are some of the determinants of prognosis (9, 10). Predictive factors such as carcinoembryonic antigen (CEA), alpha-fetoprotein (AFP), carbohydrate antigen factors 125 and 19.9 (CA125 and CA19.9) (11, 12) are also considered as prognostic indicators. In addition, tumour biology, such as presence of signet cell and BRAF mutation status, has important prognostic influence for these cases (13). Hence a careful selection of cases for CRS and HIPEC is necessary to provide a maximum survival advantage for these cases (3).
Despite being a strong indicator of survival, a specific cut-off point using PCI as an indicator has not been standardised and often when considered by itself, will not entirely predict the aggressiveness of metastases of a tumour. A clinical correlation with biochemical markers is also required, still not completely reliable in predicting prognosis (14). Small bowel region involvement has been considered more important predictor for survival than other regions, as reported in CRS-HIPEC cohort studies (13).
This is a high-volume single institution study aiming at examining the various clinical indicators and their influence on the OS for a pre-determined PCI category.
Patients and Methods
A retrospective study of a prospective dataset for peritoneal metastases from colorectal cancers was carried out in a dataset collected during the period from 2000 until September 2022. Of the total number of 1,625 cases, 415 were identified as peritoneal metastases from colorectal cancer. Cases were further grouped according to broader PCI groups of PCI<15 as (<15), PCI=15 (as 15), PCI between 16 and 20 as (16-20) and PCI>20 as (>20) and considered for analysis. Further, each individual PCI treated within the unit was also used as part of survival analysis and further discussion.
Approval was obtained from the Human Research Ethics Committee of South Eastern Sydney Local Health District, registered with number and name “2022/ETH02207: Clinical Studies in Abdominal and Peritoneal Cancers”, and conducted under the guidelines of Declaration of Helsinki statement of ethical principles for medical research.
CRS in our unit is generally performed as per the process described in Sugarbaker, 2017 (14). As part of the procedure, before surgery, the extent of peritoneal disease is measured during radiological examination and assessed during laparotomy. The PCI assessing the extent of disease is further clarified in Figure 1 (14). CRS is performed to remove all macroscopic peritoneal disease. All residual disease, including its volume and site following CRS, is recorded prospectively using the Sugarbaker’s CC-score; CC-0 is defined as, no visible peritoneal carcinomatosis after CRS; CC-1 is defined as persisting nodules <2.5 mm after CRS; CC-2 is defined as persisting nodules between 2.5 mm and 2.5 cm; and CC-3 is defined as persisting nodules >2.5 cm (15).
Sugarbaker’s peritoneal carcinomatosis score tool to determine the extent of disease.
A laparotomy was performed to assess the volume of the abdominal tumour and Sugarbaker’s Peritoneal Carcinomatosis Score tool to determine extent of disease and calculate the PCI (Figure 1). Cytoreduction was then performed on the basis that these criteria are met. The unit criteria include: 1) ability to perform a complete cytoreduction (CC-0 or 1) including the stomach, mesentery, and liver hilus; 2) adequate length of small bowel; 3) nil extra abdominal metastases. Our description of cytoreduction included total anterior parietal cytoreduction, greater and lesser omentectomy with or without splenectomy, cholecystectomy, diaphragm stripping, pelvic stripping, and abdominal viscera resection. These procedures were performed only if there was evidence of disease in these areas.
Heated intraperitoneal chemotherapy administration protocol. HIPEC is performed (in the open approach) in the context of complete or near complete cytoreduction (CC0 and CC1). Warmed to 41-46°C, the perfusate was used to achieve an intraperitoneal temperature of approximately 41.5°C. Cisplatin at 70 mg/m2, was the most common choice of chemotherapy. Enteric reconstruction is scheduled after completion of the perfusion as necessary (16).
Statistical methods. Data were analysed using IBM SPSS® software Version 24, (Armonk, NY, USA). GraphPad Prism 7® (San Diego, CA, USA) was used to create graphs, using the Kaplan–Meier analysis for OS as the primary outcome of the CRS with or without HIPEC for peritoneal metastases arising colorectal cancers, categorised according to PCI at presentation.
Mean values with corresponding standard deviation for continuous, continuous interval values and median values with corresponding range of values were determined for categorical variables and presented in the patient characteristics table and reported at the discussion. Analysis of variance of covariates including univariate and multivariate analysis was carried out using IBM SPSS® using Fisher’s coefficient for continuous variables and Pearson’s Coefficient (r) for categorical variables were used to compare the mean and medians. A p-value of 0.05 for the comparison of central tendencies was considered significant. Incidence and rate of incidence were reported for binary variables as percent, standardised to the log of 10^2.
A limit of 60 months follow-up since the index peritonectomy was set for all cases in this retrospective study. Cases not marked as dead before the last follow up date were allocated a status of alive. The last follow up date for cases was included in the survival calculations and was marked as “lost to follow up” at that time point and censored. Microsoft EXCEL® (Redmond, WA, USA), was used to determine the final status and lost to follow up cases including the survival proportion and probability of survival over time.
The Cox regression method for proportional hazard ratio was used to measure proportion of survival and the Kaplan–Meier survival curves were used to display survival probability over time.
Results
The population of the current study was grouped into PCI of <15, 15, 16-20 and >20. Out of 340 cases in the <15 group, 137 (40%) were males. Eight out of 15 (53.3%) in the PCI 15 group were males (Table I). Whereas 14 (42.42%) of 33 and 18 (66.67%) of 27 were males in the 15-20 and >20 groups, respectively. Mean age for the <15 group was 56.97 [Std Dev (SD)=13.63], for PC 15 group was 58.08 (SD 13.07), for 15-20, 51.67 (SD 14.96) and 55.15 (SD 13.67) for the >20 group. Similarly, median age was 58 (range=18-123 months), 63 (range=34-75 months), 54 (range=16-79 months) and 55 (range=29-80 months) for the groups of <15, 1, 16-20, and >20, respectively.
Patient characteristics categorised into major peritoneal cancer index (PCI) groups of PCI <15, 15, 16-20, and >20.
Average PCI (Table I) was 6.9 (SD 3.94), 17.73 (SD 1.38) and 28.44 (SD 4.27) for the <15, 15-20 and >20 categories. Median morbidity grade in all the groups was similar at 2 with a range of (0-5) for <15 and (0-4) for rest of the groups. 73.61% (n=251) recorded a score of 0 to 2 for the <15 group, 66.67% (n=10) for 15, (63.64% (n=21) for 16-20 and 59.26%, (n=16) for >20 groups. The median CC-score was 0 for all groups. PCI <15 and PCI-15 had 99.12% and 100% complete cytoreduction, respectively, followed by 93.94% and 85.19% achieving CC0 for the groups of PCIs 16-20 and >20, respectively.
HIPEC was administered in 87.65%, 93.33%, 84.85% and 74.07% cases in the PCI <15, 15, 16-20 and >20 groups. Both mitomycin and oxaliplatin were equally preferred with 51.34% receiving mitomycin in the <15 group and 50% in the 15 group. However, oxaliplatin was preferably used in the higher PCI groups of 16-20 and >20 categories at 54% and 55%, respectively (Table I).
The mucinous cases accounted for 20.29% in the <15 category. In contrast, mucinous cases were 53.33%, 57.58%, 62.96% in 15, 16-20 and >20 groups (Table I). The difference in mucinous prevalence between the groups was significant (p<0.0001) (Table II).
Cox Regression analysis of survival for variables between peritoneal cancer index groups.
A further linear regression model for variables between groups revealed significant difference for morbidity (p=0.001), CC-score (p<0.0001), CA125 (p<0.0001) and CA199 (p<0.0001) (Table II).
A Pearson’s coefficient correlation was performed for incidence of death and a two tailed test of significance calculated between groups. PCI 15, 16-20 and >20, groups displayed a negative correlation with PCI <15, and a 2-tail significance of (p=0.012) between the <15 and >20 groups indicating the strongest correlation between groups. The rest of the groups showed a positive correlation with none displaying a significant difference in incidence (Table III).
Pearson’s correlation analysis between peritoneal cancer index groups.
The Log-rank hazard ratio (HR) for 5-year survival using the Cox regression analysis over time (t) and the significance between the groups using the Log-rank (Mantel–Cox) test were: HR=2.121 and p=0.0338 for <15 vs. 15, HR=2.748 and p<0.0001 for <15 vs. 16-20, HR=3.158 and p<0.0001 for <15 vs. >20, HR=1.262 and p=0.5658) for 15 vs. 16-20, HR=1.566 and p=0.2771 for 15 vs. >20 and HR=1.204 and p=0.5355 for 16-20 vs. >20 (Table IV).
Hazard ratio using the Cox regression analysis for proportional hazard and corresponding p-value between peritoneal cancer index (PCI) groups.
Median survival for the categories of <15, 15, 16-20, and >15 was 43.967 (95%CI=28.31-59.63), 20.67 (95%CI=5.01-36.33), 19.50 (95%CI=3.84-35.16), and 14.30 months (95%CI=1.36-29.96), respectively (Figure 2, Table V).
Kaplan–Meier survival curve comparison between peritoneal cancer index (PCI) groups.
Median survival for the peritoneal cancer index (PCI) groups.
A survival function pattern (Figure 3) was emulated using the Cox regression analysis with separate function curve for each PCI. There was a predicted gradual decrease in slopes per PCI score with PCI scores of 16 [n=8, median survival (OS)=9.2], 24 (n=4, OS=29.3), and 29 (n=3, ms=8.43), showing a regressed median survival as compared to their corresponding higher PCI. Factors such as low population cohort within these groups and individual clinical indicators may need to be considered to help derive a definitive conclusion.
A Kaplan–Meier survival function pattern for individual peritoneal cancer indexes (PCIs) treated within the cohort with individual survival proportional to population in each group and respective median survival.
A cox regression analysis for survival for CRS-HIPEC and CRS only cases revealed a survival advantage for HIPEC (median survival=39.8, n=357) vs. non HIPEC (median survival=24.4, n=54), (p=0.005) with an HR for OS for HIPEC of HR=1.63 (95%CI=1.60-1.66) (Table VI).
Hazard ratio for overall median survival for the hyperthermic intraperitoneal chemotherapy (HIPEC) group vs. no HIPEC group and chemotherapy agents within the HIPEC group.
Further, the cox regression model for proportional hazard ratio over time applied to the entire cohort between mitomycin vs. oxaliplatin revealed a better OS with a median survival of 56 months for the oxaliplatin group versus 33 months for the mitomycin group (p=0.002) (Figure 4) and an HR for OS of 1.70 (95%CI=1.66-1.73) (Table VI).
A Kaplan–Meier survival function pattern using Cox Regression analysis for the proportional hazard ratio for Mitomycin vs. Oxaliplatin used for the cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) group.
Discussion
In a multivariate analysis, age and sex did not significantly affect survival (p=0.701 and p=0.327) in all the groups, with females being highly represented in <15 and 16-20 groups at 59.41% and 57.58%, respectively, and males with higher representation in the 15 and >20 groups at 53.33% and 66.67%. Age was equally distributed between the groups with highest median age of 58.08 years for the PCI 15 group (Table I).
Our study revealed a median morbidity grade of 2 across the groups and was significant for survival at p<0.0001 (Table I). There was a significant difference in distribution between the groups for morbidity as well, (p=0.001) (Table II).
Predictively, PCI has been a strong predictor of survival for CRC (p<0.0001). Mitomycin was preferred for the <15 group at 51.34% as compared to oxaliplatin (48.32%) and was reversed as a preference for higher PCI groups.
There was no sign of difference between oxaliplatin and mitomycin, with an overall effect on survival at p=0.099 between the groups revealing nil significance for survival. Similarly, tumour markers, AFP (p=0.457), CA125 (p=0.439) and Pre CEA (p=0.136) did not display a significant difference between the groups with CA199 showing a significance at multivariate p-value of 0.021. However further linear analysis of the values for tumour markers revealed significant differences in distribution for CA125 (p<0.0001) and CA199 (p<0.0001) between the groups.
In an early study by Pestieau et al., the overall comparison made between PCI<10, PCI 11-20, and PCI >21 groups indicated a median survival of 48, 24 and more than 12 months, respectively, (p<0.0001) between the groups (1). The PCIs were grouped differently in this study, however, the median OS of 14.3 months for the >20 cohort was similar to the >21 group in that study (Table V).
Vaira et al., reported that the OS in the PCI less than 16 and PCI greater than 16groups did not differ significantly (p<0.290). However, the same study reported a significance in OS between the groups of CC-score 2 and 0 (p<0.0001), (5) suggesting that CC-score influences OS. Interestingly in the current study, although there was a difference in the CC-score between the groups (p<0.0001, Table II), there was no major difference in survival for CC-score between the groups with a multivariate p-value of 0.113 (Table I).
Hallam et al. and Brandl et al., predicted the histology of the primary tumour influencing survival in CRC (9, 10). In addition, Brandl et al. reported in a comparison of mucinous carcinoma, described by abundance of mucous in tumours, to that of other carcinomas in CRC, that the PCI did not have much influence on the OS for cases with mucinous cancers (PCI <16 vs. PCI ≥16, p=0.935) as compared to those with non-mucinous cancers (PCI <16 vs. PCI ≥16, p=0.009) (10). The present study demonstrated similar results between mucinous and non-mucinous cases. For instance, although the incidence of mucinous presentation was progressively higher in higher PCI groups, 20.29% for the <15 groups and 53.33%, 57.58%, 62.96% in 15, 16-20, and >20 groups the comparison of the incidence between the groups (p<0.001) (Table I) did not reveal significant differences in survival outcome between the groups for this factor; the multivariate analysis revealed a p-value of 0.672 and the cox regression multivariate analysis revealed a p-value of 0.079 (Table II). Presence of Signet and Goblet cells, however, will influence OS (10) which was deemed out of scope for this study.
Faron et al. have discussed the perfect linear relationship of PCI and OS and disproved a cut-off based on PCI value only for CRS HIPEC (19). This has been depicted in the survival function plot using the cox regression analysis for each PCI within the cohort (Figure 3) in the current study showing that almost in all cases that the survival curves are in order of PCI. With little difference for PCI 4 till 9 where the 5 year survival was 50% and PCI 10 till 20 where the 5 year survival was nearly 20%. These results suggest that with careful case selection, PCIs as close as 20 can have long term survival.
A Pearson coefficient correlation for mortality within 60 months, revealed a negative correlation between <15 to rest of the groups with PCI 15, 16-20 and >20 with a significance between the the <15 and >20 groups (p=0.012). A significant difference in OS was confirmed between PCI<15 and 16-20 (p<0.0001), <15 and >20 (p<0.0001), and <15 and 15 (p=0.0338) suggesting better prognosis for cases with PCI <15 (Table III and Table IV).
Similarly, a Kaplan–Meier survival curve between the groups also revealed a survival advantage for the <15 group compared to rest of the PCI groups with an overall significance of (p<0.0001) (Figure 2).
A χ2 test for difference (Δ) of median survival in months for individual PCIs to overall median survival for <15 group showed an expected survival advantage for PCIs less than 15 versus PCIs more than 15. Interestingly however, taking observational power of numbers in consideration and excluding PCIs with less than 3 cases out of the analysis, some PCIs >15 such as, PCI 17, 18, 19, 21 and 22 with values of (n=8, Δ=1.56, χ2=0.12, 95%CI=0.02-0.25), (n=7, Δ=2.66, χ2=2.66, 95%CI=0.34-0.48), (n=5, Δ=11.43, χ2=6.32, 95%CI=6.19-6.46), (n=4, Δ=8.63, χ2=3.60, 95%CI=3.47-3.74) and (n=3, Δ=11.65, χ2=6.57, 95%CI=6.43-6.70), respectively, showed better survival compared to PCI 15 (Table VII).
Difference of overall median survival for selected peritoneal cancer index (PCI)s >15 that has a better survival to PCI=15.
In a study conducted earlier in the same institution as this study, the median OS for CRC cases with low PCI (PCI <5) receiving CRS and intraperitoneal chemotherapy intraoperatively was 80.8 months (8).
In the stratified survival function pattern for all PCIs in the current study, PCIs 2 and 3 did not reach median at 60 months with PCIs 1, 4 and 5 reaching a median survival of 52.97 (n=11), 54.1 (n=31) and 55.9 (n=26) (Figure 3). In addition, it is evident that the survival curves are in order of PCI with more than 40% chance of 5-year survival for PCIs 5 till 9 and a 20% chance of 5-year survival for PCIs 9 till 20 (Figure 3).
Some PCIs showed regressed overall survival (OS) compared to their consecutive higher values, PCI 6 [OS=38.63, n=36] vs. PCI 7 (OS=Undefined/not reached at 60 months, n=22) and PCI 9 (OS=42.43, n=23), PCI 13 (OS=25.3, n=14) vs. 14 (OS=33.33, n=18). With PCI 9 performing better that PCI 6 in our study, this contrasts with a recent study published by Livin, 2022 (17) suggesting a cut off PCI 7 (Figure 3).
A similar pattern also was observed for PCI 16 (OS=9.2, n=8) vs. PCI 17 (OS=22.23, n=8). In fact, PCI 16 has been one of the worst performers compared to its neighbouring PCI groups. Also, PCI 27 (OS=3.4, n=1) vs. PCI 28 (OS=32.32, n=3), PCI 29 (OS=8.43, n=3) vs. PCI 30 (OS=Undefined, n=3) and PCI 31 (OS=14.37, n=3) and PCI 32 (OS=11.1, n=2) showed similar regression in survival. However, factors such as low numbers, less observed power within groups and other clinical indicators such as BRAF, KRAS, and MMR status (13) including other metastatic sites, also need to be considered for individual scenarios before drawing conclusions (Figure 3).
There is a high correlation between CRS and HIPEC and high morbidity in the literature, yet the 30 day mortality is low and a good survival outcome for selected cases help alleviate the concerns presented during the hospital stay postsurgery (2, 18, 19). Similarly in the current study, there was nil significance for mortality between the modalities of CRS-only and CRS and HIPEC groups with a p-value of 0.522. A comparison using the cox regression model for proportional HR over time, between the CRS HIPEC group and CRS only group in the overall population, revealed a better OS with a median survival of 39.8 months for the overall CRS HIPEC group (57% survival at 60 months, n=360) as compared to 24.4 months for the CRS only group (33% survival over 60 months, n=55) (p=0.005) (Figure 5).
A Kaplan–Meier survival function pattern using Cox Regression analysis for the proportional hazard ratio for cytoreductive surgery (CRS) vs. cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) groups.
Further, with a HR=1.70 (95%CI=1.01-2.38, p=0.005) the cox regression model for proportional hazard for survival favoured oxaliplatin with a better median OS of 56 months for the oxaliplatin group (60% survival at 60 months, n=175) versus 33 months for the mitomycin group (53% survival at 60 months, n=176) (p=0.002) (Figure 4).
A similar survival function pattern using the Cox Regression analysis for proportional hazard was observed for the CRS with HIPEC group within the cohort. This highlighted, more than 60% 5-year survival for PCIs 1 till 7 and more than 35% 5-year survival for PCIs 8 till 17. And within the higher PCI range, 50% of the population with PCIs 24, 28, 30 reached survival of more than 30 months (Figure 6). Suggesting that there is no one PCI value that can be used as a cut off but there is a progressive fall in long term survival with increasing PCI keeping in line with the results presented by Kamada et al. (20); and that perhaps 20 would be a more appropriate cut off.
A Kaplan–Meier survival function pattern for individual peritoneal cancer indexes (PCIs) treated within the cytoreductive surgery (CRS) with complete cytoreduction (CC0) and hyperthermic intraperitoneal chemotherapy (HIPEC) group with individual survival proportionate to the population in each group and their respective median survival.
Conclusion
PCI is one of the strongest predictors of survival in patients with peritoneal metastases from colorectal cancers as it was confirmed by the linear correlation to survival in this study. However, whilst PCI is so important for prognosis in CRC and gastric cancer, data for both low and high grade appendiceal cancers and mesothelioma have entirely different outcomes and we have no PCI limits for these tumours.
Assessment of PCI using both laparoscopic and radiological procedures is imperative to help optimal planning. Detecting synchronous peritoneal metastases early and prompt treatment can help prevent recurrence and increase survival. Small bowel involvement, pre-operative performance status, BRAF mutation status including presence of signet cell histology should also be part of the treatment decision process. The combination treatment of cytoreduction and hyperthermic intraperitoneal chemotherapy with an appropriate choice of chemotherapy agent, will increase survival for these cases. The current study was conducted at a single institution. A multiple institution study using the same variables could help improve observational power of the study.
Footnotes
Authors’ Contributions
DLM conceived and conceptualised the study, SB designed and performed the literature review, data collection and analysis, preparation of the figures and initial drafting of the manuscript. Further review, critical assessment done by DLM and RW. All Authors read and approved the final manuscript.
Conflicts of Interest
The Authors have no conflicts of interest to declare in relation to this study.
- Received May 24, 2023.
- Revision received June 23, 2023.
- Accepted June 26, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.