Tumor marker based survival analysis for patients with pseudomyxoma peritonei of appendiceal origin: A retrospective cohort study


 Background Pseudomyxoma peritonei (PMP) is a rare disease, the prognosis of overall survival (OS) is affected by many factors, present study aim to screen independent prediction indicators for PMP and establish prediction model for OS rates in PMP.Methods 119 PMP patients received cytoreductive surgery (CRS) combined with hyperthermic intraperitoneal chemotherapy (HIPEC) in our center for the first time were included between 01/06/2013 and 22/11/2019 . The log-rank test was used to compare the OS rate among groups, subsequently, variables with P<0.10 were subjected to multivariate Cox model for screening independent prediction indicators. Finally, the prediction models for OS in PMP will be established.Results Univariate analysis showed that Barthel Index Score, albumin, D-dimer, CEA, CA125, CA19-9, CA724, CA242, PCI, degree of radical surgery, histopathological grade were signiﬁcant predictors for OS in PMP. At multivariate analysis, sex, D-dimer, CA125, CA19-9, and degree of radical surgery were independently associated with OS rate in PMP. ROC analysis was performed to calculate discrimination ability of prediction model and the area under curves (AUC) was 0.902 (95%CI: 0.823-0.954). Finally, nomogram was plotted by the independent predictive factors for PMP.Conclusions Several factors (sex, degree of radical surgery, D-dimer, preoperative CA125 and CA19-9) have independent prognostic value for survival in PMP, the tumor based prediction model has better prediction value, more researches are need to verify and improve the prediction model.


Abstract
Background Pseudomyxoma peritonei (PMP) is a rare disease, the prognosis of overall survival (OS) is affected by many factors, present study aim to screen independent prediction indicators for PMP and establish prediction model for OS rates in PMP.
Methods 119 PMP patients received cytoreductive surgery (CRS) combined with hyperthermic intraperitoneal chemotherapy (HIPEC) in our center for the rst time were included between 01/06/2013 and 22/11/2019 . The log-rank test was used to compare the OS rate among groups, subsequently, variables with P<0.10 were subjected to multivariate Cox model for screening independent prediction indicators.
Finally, the prediction models for OS in PMP will be established.
Results Univariate analysis showed that Barthel Index Score, albumin, D-dimer, CEA, CA125, CA19-9, CA724, CA242, PCI, degree of radical surgery, histopathological grade were significant predictors for OS in PMP. At multivariate analysis, sex, D-dimer, CA125, CA19-9, and degree of radical surgery were independently associated with OS rate in PMP. ROC analysis was performed to calculate discrimination ability of prediction model and the area under curves (AUC) was 0.902 (95%CI: 0.823-0.954). Finally, nomogram was plotted by the independent predictive factors for PMP.
Conclusions Several factors (sex, degree of radical surgery, D-dimer, preoperative CA125 and CA19-9) have independent prognostic value for survival in PMP, the tumor based prediction model has better prediction value, more researches are need to verify and improve the prediction model.

Background
Pseudomyxoma peritonei (PMP) is a rare disease characterized by disseminated mucinous ascites within peritoneal cavity which most often originating from perforated appendiceal epithelial neoplasma.
[1] Smeenk et al. [2] estimated the incidence of PMP is about two per million annually from the Netherlands, other major reseach centers suggests that the actual incidence may be higher at 3-4 operable cases per million per year. [3] Cytoreductive surgery (CRS) combined with hyperthermic intraperitoneal chemotherapy (HIPEC) is recommended as the optimal treatment for patients with PMP, [4] the recurrence rate has been obviously decreased than before, oppositely, the overall survival (OS) rate improved greatly.
Although the long-term outcomes after treatment are impressive for patients with PMP, there's still a signi cant recurrence of the disease. [5,6] It has been con rmed that there are many factors related to the prognosis of PMP, for instense, sex [7], extent of previous surgery [8], histopathological grade of tumor, tumor marker levels, [9] degree of radical surgery, [1] and so on, among of tumor markers, CA19-9, CEA, and CA125 had been widely veri ed. [10] Although several demonstrated factors could affect the prognosis of patients with PMP, to our best knowledge, there are very few studies had established prediction model for PMP patients. In present study, we want to reevaluate whether all the above traditional factors have predictive value for PMP patients, subsequently, we intend to evaluate the predictive value of new tumor markers (CA724 and CA242) for PMP. On this foundation, a tumor markers based model to predict the prognosis of PMP will be established, which may be helpful for the prognosis judgment and treatment intervention in PMP patients .

Patients
The ethics committee of Peking University Aerospace School of Clinical Medicine approved of present study, all patients signed informed consent before CRS and consented to be followed up after surgery.
We retrieved the diagnostic name of'pseudomyxoma peritonei'in the special follow-up database from Peking University Aerospace School of Clinical Medicine between 01/06/2013 and 22/11/2019, a total of 886 patients with PMP diagnosis were acquired. In order to ensure the reliability of the research, 734 patients with PMP whose operation were performed not in our center were excluded from present study, the detailed reasons were as follows, for rst, different hospitals may use different instruments or methods to detect tumor markers, which cannot guarantee the consistency of test results; secondly, although CRS combined with HIPEC are the optimal treatment for PMP patients, we found that there were still many patients who had only undergone CRS or chemotherapy (intravenous or intraperitoneal) in nonspecilist hospitals, PMP treatment is best in an inter professional team approach including specialists, oncology trained specialty nursing, and when necessary, pharmacists, collaborating for optimal patient care and outcomes, [5] therefore, the above 734 subjects were ruled out.
Thus, the 152 remaining cases received CRS and HIPEC treatment in our center for the rst time were included. Histopathological results of resected specimens were interpreted by two experienced pathologists according to the WHO 2010 classification, which were categorized into low-grade appendiceal mucinous neoplasms (LAMNs) and mucinous adenocarcinomas (MACAs). [11] Follow up protocol All patients were routinely followed up every 3 to 6 months, tumor markers (such as CEA, CA19-9, CA125, CA72-4 and CA242) and enhanced computed tomography (CT) of abdominopelvic were routinely examined, If any discomfort occurs during discharge, the patient should return to the hospital at any time. If the patient does not return to our hospital for further consultation, we will follow up the patient by telephone to record the the patient's examination results.

Endpoint event determination
Due to the limitation of medical conditions in china, most patients with PMP missed the opportunity of early operation for complete CRS (CCRS), so a large proportion of whom were performed maximal tumor debulking surgery (MTD), the endpoint event was the death of PMP patients.

Tumor markers determination
All tumor markers were determined within 7 days. All marker measurements were performed according to manufacturer instructions, CEA (ng/ml), CA125 (U/ml), and CA19-9 (U/ml) were measured by chemiluminescence immunoassay (CMIA) (Abbott, America), the same to CA724 (U/ml) (Autobio, China), while CA242 (kU/L) was tested by ow uorescent technology (Luminex, America). Internal Quality Control (IQC) was performed for all 5 tumor markers before testing, at the same time, we also participated in the External Quality Assessment (EQC) twice a year.
peritoneal carcinomatosis index (PCI) The PCI scoring system divides the abdomen into nine anatomical areas with four further areas of the small bowel. Tumor is assessed in each area and a score of 0-3 is given for each of the 13 areas (0 for no tumor, 1 for nodules <0.5cm, 2 for nodules between 0.5 and 5cm, and 3 for nodules >5cm). The total score is then calculated by adding all the scores, and ranges from 0 to 39 [12].

CRS and HIPEC
The CRS of PMP was performed consistent with standard operation method, complete removal of all visible disease is scored as CC0 cytoreduction and residual disease less than 0.25cm is scored as CC1, CC0 and CC1 are considered CCRS. If the patient cannot achieve complete cytoreduction, debulking treatment would be performed, any residual tumor deposit between 0.25 and 2.5cm is scored as a CC2 cytoreduction while residual tumor deposits >2.5cm are scored as CC3 cytoreduction, CC2 and CC3 are considered as having MTD. [13] Once cytoreduction is complete, intra-operative hyperthermic chemotherapy is delivered. 5-uorouracil (5-Fu) (1000mg) together with Cisplatin (80mg) heated to 43℃ and continuously infused using a HIPEC machine for 1h.

Statistical Analysis
Statistical analysis were performed by SPSS (Version 16.0), MedCalc (Version 15.2.2), X-Tile (3.6.1), and R Software (3.6.2). All continuous data will be compared by using t test or ManneWhitney U test, as appropriate. Pearson's c 2 test or Fisher's exact test, where appropriate, was used for analysis of categorical data. The Kaplan-Meier method and log-rank test were used to compare the OS rate among groups, afterwards, variables with P<0.10 were subjected to multivariate Cox models, Cox proportional hazards models were used to calculate the hazard ratio and 95% confidence interval (CI). A nomogram was plotted by R software to facilitate risk assessment for PMP. Two sided p values < 0.05 were considered statistically signi cant.

Results
Patient demographics 152 PMP subjects underwent CRS and HIPEC for the rst time in our center, one patients of sigmoid colon origin and four died after CRS due to serious infection during hospitalization were excluded from this study. During the follow up period, 15 patients lost of follow up, afterwards, 132 patients were followed up, among of whom, whose follow up time less than 6 months were also excluded (n=13), ultimately, 119 PMP patients were included in present study, study schematic was shown in Figure 1.
In order to avoid bias in the study population as much as possible, comparative analysis of the baseline data was performed between the included (n=119) and excluded (n=33) subjects, there was no signi cant difference in sex ratio, age, PCI, Barthel Index Score, and the degree of radical operation between the two groups (all P>0.05), however, different proportion of histopathological grading was found between the two groups (P<0.05) ( Table 1)

Impact of independent variables on patient survival
The ability to parse tumors into subsets based on biomarker expression has many clinical applications, many former studies employed the upper limit of reference range of tumor markers as the best cut-point. In 2004, Camp, R. L. [14] reported X-Tile plot could serve as a new bio-informatics tool to visualize the best cut-points for creating such divisions. In this study, X-Tile software was used to calculate the best cutpoint of continuous variables (age, Barthel Index Score, albumin, D-dimer, CEA, CA125, CA19-9, CA724, CA242, and PCI ) in independent variables, however, we did not calculate the cut-off value for hemoglobin, because the hemoglobin level for anemia diagnosis in female and male is different (110g/L for female and 120g/L for male).
At univariate analysis, Barthel Index Score, albumin, D-dimer, CEA, CA125, CA19-9, CA724, CA242, PCI, degree of radical surgery, pathology were all significantly associated with OS rate in PMP. Although sex factor did not meet the criteria for inclusion in multivariate analysis, literature reported women tend to present at an earlier stage than men [1], we speculate that sex has a great in uence on the prognosis of PMP, ultimately, sex factor was also included into Cox regression analysis. At multivariate analysis, sex, D-dimer, CA125, CA19-9, and degree of radical surgery were independently associated with OS rate in PMP. (Table 2). Cox regression analysis generated variables including SUR and XBE, afterwards, we calculated new variables risk based on generated variables, the formula was as follows: risk=1-SUR**EXP (XBE), according to the risk variable, ROC analysis was performed to calculate discrimination ability of prediction model, the area under curves (AUC) was 0.902 (95%CI: 0.823-0.954) (Figure 3). Finally, the independent predicting factors were used for drawing nomogram for PMP (Figure 4).

Discussion
PMP is a rare disease, which tends to be an incidental nding either on imaging or during exploratory surgery performed for other indications. In present study, we demonstrated several factors for predicting OS rate of PMP, such as degree of radical surgery and so on, subsequently, we established prediction model for OS in PMP. Cox proportional hazard regression model showed that male, MTD, an increase of preoperative CA125 and CA19-9 level, and elevated D-dimer level were independently associated with poor survival for PMP.
Among the three commonly used tumor markers for PMP [15], CA19-9 seems to be optimal independent prognosticators for PMP, which not only could predict survival but also predict recurrence, a lot of researches con rmed this conclusion [16][17][18][19]. CA125 can predict ovarian cancer in general practice [20], which is also expressed in peritoneal malignancy, and can be elevated in patients with any source of peritoneal irritation [1]. In present study, CA125 seems also to be a usefull marker for prediction survival of PMP, researches with larger sample size and longer follow-up time are needed to verify this conclusion. Although univariate analysis revealed elevated CEA level was associated with worse survival in PMP, nevertheless, multivariate analysis did not reach a signi cant statistics, former study also con rmed CEA owns low value in prediction survival of PMP [16]. Present study also evaluated CA724 and CA242 in PMP patients, univariate analysis all showed elevated levels of the two markers were all associated poor survival of PMP, while in multivariate analysis, CA724 did not show signi cant prognostic value for survival of PMP, while CA242 was closed to statistical signi cant, more researches are needed to a rm it.
Completeness of cytoreduction is one of the most important prognostic factors for PMP [1], the present study revealed that MTD subgroup had a obvious poor survival than the CCRS group, this result was similar to the previous studies [21,22], differently, a large proportion of participants in the former study can reach CCRS, while in our study, the majority of patient can only undergone major debulking surgery, we speculate that most PMP patients in China cannot get correct PMP diagnosis timely and receive standard treatment. This study found that male of PMP had a low level OS rate than female, which is consistent with previous research reports [2,7], women tend to present at an earlier stage than men for secondary to the rapidly enlarging ovarian masses, which become symptomatic or are obvious clinically [1], so women can get more timely treatment. The extent of disease is assessed by the PCI, a PCI≥20 always representing unresectable disease,[23] former studies con rmed that PCI was the risk factors for postoperative morbidity in univariate analysis, however, no statistical signi cant correlation was found during the multivariate analysis.
[24] Present study found that a high D-dimer level was associated with a poor survival for PMP, to our knowledge, this was the rst study to evaluate the prediction value of Ddimer in PMP, in the future clinical practice, clinicians should pay more attention to the D-dimer level of in PMP patients.
A part of researchers reckon that the prognosis of PMP correlates closely to histopathological classi cation,[25] ,oppositely, different opinions suggest that PMP is unlike other tumors, histopathology does not reliably predict tumor behaviour, including the likelihood of recurrence.
[26] Present study revealed a high grade histopathology denotes a poor survival for PMP in univariate analysis, while in multivariate analysis, no correlation between histopathological grade and OS rate was found, so we speculate that predictive value of histopathological grade for PMP is relatively small. Similarly, although Barthel Index Score and albumin seemed to associated with prognosis of PMP in univariate analysis, which was similar to the former study,[27,28] the multivariate analysis did not reach signi cant difference, Barthel Index Score re ect the ability to perform the activities of daily life[29] and albumin re ect nutrition condition in PMP, which suggest that those effects on the prognosis of PMP patients is limited, but this conclusion needs to be con rmed by a large number of studies.
The present study established prognostic model for OS in PMP, the discrimination ability of OS in PMP was 0.902, so we think this is a valuable prediction model for PMP, which could provide more help for the prognosis judgment and treatment intervention of PMP patients, however, more studies are needed to con rm the conclusion, in particular with prospective large sample studies.
There were several limitations of present study. First, due to the limitation of retrospective study, some data have not been followed up. Secondly, because the survival time of PMP patients is signi cantly longer than before, the proportion of end-point events is relatively small, it may lead to the instability of statistical conclusions, therefore, a longer follow-up is needed to con rm the conclusions in the future. Finally, although the prediction model is relatively good for PMP, we think that there is still some other factors not included into the prediction model, such as KRAS mutations, which had been proved independently prognostic for progress free survival (PFS) in PMP patients, we speculate that KRAS mutations may also be independent prognostic factors for OS in PMP, future work will further verify this hypothesis.

Conclusion
To conclude, several factors (sex, degree of radical surgery, D-dimer, preoperative CA125 and CA19-9) have independent prognostic value for OS rate in PMP, the prediction model based on the above factors has better prediction value, more researches are need to verify and improve the prediction model.   Figure 1 Study schematic. A total of 886 patients with PMP was retrieved , 734 patients not undergone CRS+HIPEC in our center for the rst time were excluded the remaining 152 patients undergone CRS+HIPEC were included, patients of sigmoid colon origin (n=1) and died during hospitalization (n=4) were excluded. Patients lost of follow up (n=15) and whose follow up time less than 6 months (n=13) were also excluded, ultimately, 119 PMP patients were included in present study. PMP-pseudomyxoma peritonei; CRS-cytoreductive surgery; HIPEC-hyperthermic intraperitoneal chemotherapy.

Figure 2
Overall survival curve of 119 PMP patients Nomogram for prediction of overall survival rate in PMP patients