Survival benefits with EPIC in addition to HIPEC for low grade appendiceal neoplasms with pseudomyxoma peritonei: a propensity score matched study

Introduction

Appendiceal cancer is a rare malignancy with a rising incidence, occurring in roughly 1.2 per 100,000 per year and making up 0.4% of all gastrointestinal malignancies [1], [2], [3]. Diagnosis is often made incidentally, either during surgery for other causes or histologically after removal for presumed appendicitis [3]. Classification is by histological subtyping, with mucinous appendiceal neoplasias including low and high grade appendiceal mucinous neoplasms (LAMN and HAMN) as well as adenocarcinoma [2, 4]. Prognosis varies depending on the subtype, stage, histological grading, mucin deposits outside of the right lower quadrant as well as the presence of intracellular mucin [5]. Pseudomyxoma peritonei (PMP) is a clinical syndrome of intraperitoneal accumulation of mucus due to mucinous neoplasia, most commonly arising from appendiceal neoplasia [4, 6]. PMP is split into low-and high-grade dependent on its histological features, also known as disseminated peritoneal adenomucinosis (DPAM) or peritoneal mucinous carcinomatosis (PMCA) for low and high-grade respectively [4].

LAMNs with PMP responds poorly to systemic chemotherapy and treatment historically consisted of debulking or drainage of mucinous disease alone, with a poor 10-year survival between of 32% [7]. Cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (HIPEC) introduced by Sugarbaker [8] significantly improved the survival of these patients with studies showing a 10 year survival of 63% [7]. Early postoperative intraperitoneal chemotherapy (EPIC) was introduced over 30 years ago [9] and is an alternative and additional form of perioperative intraperitoneal chemotherapy. The use of EPIC compared to HIPEC declined due to concerns of increased morbidity and survival benefits of HIPEC over EPIC, as well as the increased resource utility required to deliver EPIC, often requiring patients to stay in intensive care unit (ICU). A systematic review performed showed mixed data, with nine of thirteen studies showing increased morbidity with EPIC (either alone or in combination with HIPEC) and those in direct comparison of HIPEC to HIPEC + EPIC, five out of seven studies showed survival benefit with EPIC addition [10].

In those that showed survival benefit [11, 12], there were criticisms proposed regarding selection bias in that patients were younger and less comorbid when selected for additional EPIC [10]. Ideally, a randomised control trial (RCT) should be performed to compare this and minimise bias. Given the significant resources involved in an RCT, techniques have been developed to mimic RCTs in non-randomised data. One of these strategies is propensity score matching, in which observed covariates are weighted using a logistic regression model to generate a score. This score is then matched between the control and treatment group, minimising bias in non-randomised samples to allow comparison of cohorts [13].

Using propensity score matching, we aim to examine overall survival (OS) for patients with low grade appendiceal cancer and peritoneal carcinomatosis being treated with peritonectomy and HIPEC, with and without the addition of EPIC between matched cases.

Materials and methods

All patients with LAMN and DPAM who underwent cytoreductive surgery at St George hospital between January 1996 and August of 2020 were included in this retrospective analysis. Patients who were undergoing redo-peritonectomy were excluded.

Results

There were 224 patients identified, 124 of which were female (55.4%). Of these, 52 (23.2%) underwent HIPEC alone and 172 (76.8%) HIPEC + EPIC. Table 1 shows the pre-propensity score matching clinical data and demographics between the HIPEC and HIPEC + EPIC group. There were significant differences between the HIPEC and HIPEC + EPIC groups in age (58.4 vs. 52.2 years, p=0.001), median CC score (1 vs. 0, p=0.014), choice of HIPEC agent (p<0.001) and ASA status (<0.001). There was no significant difference between the groups in either median grade of morbidity complication or frequency of major morbidity (p=0.495 and 0.492, respectively).

Clinical characteristics and demographics between the post-propensity score matching HIPEC and HIPEC+EPIC groups are shown in Table 2. Similar to the pre-propensity matching, in comparing the HIPEC and HIPEC + EPIC groups there is significant difference noted in age (58.4 vs. 54.3, p=0.044) and choice of HIPEC agent (p=0.045). There are additional significant differences noted in CA19.9 (66.8 vs. 101.5, p<0.001), CEA (43.4 vs. 31.3, p=0.045) and total length of stay in days (23.5 vs. 25.0, p=0.028). There remained no significant difference in median morbidity grade or frequency of major morbidity.

Non-parametric testing of the clinical characteristics in relation to survival pre-propensity matching is shown in the left-hand side of Table 3. There was no statistical significance of impact on survival for sex, age, choice of HIPEC agent or length of stay in ICU. There was evidence of effect on survival noted with univariate analysis for PCI (p<0.001), morbidity grade (p=<0.001), CC score (p=0.003), ASA status (p=0.017), total length of stay (p=0.001) and tumour markers; CA19.9 (p<0.001), CA125 (p<0.007) and CEA (p=0.001). Multivariate analysis was statistically significant in the same categories, with the addition of choice of HIPEC agent (p=0.003) not present in univariate. Pearson correlation coefficients revealed positive correlation with all factors, with the strongest correlation seen in PCI at 0.376.

Post-matching analysis of variables in relation to survival is shown in the right-hand side of Table 3. The following impacted survival with univariate analysis for PCI (p<0.001), morbidity grade (p=0.001), ASA status (p=0.018), total length of stay (p=0.001) and tumour markers; CA19.9 (p=0.001) and CEA (p<0.023). CC score, choice of HIPEC agent and CA-125 no longer showed impact on survival. In addition, ICU length of stay had evidence of impact on survival (p=0.026). All factors with evidence of impact on survival were positive correlated, with the strongest relationship once again demonstrated by PCI score at 0.390. Age did not impact survival in pre- or post-propensity matching.

The addition of EPIC to HIPEC was associated with stronger survival in hazard ratios (HR) at all time points measured (Table 4). This effect increased at each interval, from a HR of 1.12 (95% CI 1.07–1.16) at six months, to a HR of 1.37 (95% CI 1.33–1.42) at 6 months.

Pre-and post-propensity matching Kaplan–Meier curves for survival are shown in Figures 1 and 2. Pre-propensity matching median survival (Figure 1) was 92.97 months in the HIPEC group, with HIPEC + EPIC not reaching a median with more than 50% surviving at time of analysis (p<0.001). Post-propensity matching median survival (Figure 2) was significantly different between the HIPEC and HIPEC + EPIC groups at 92.97 vs. 127.3 months, respectively (p=0.021).

Figure 1:

Kaplan-Meier curve of survival for patients treated with HIPEC or HIPEC + EPIC pre-propensity matching. HIPEC, hyperthermic intraperitoneal chemotherapy; EPIC, early postoperative intraperitoneal chemotherapy.

Figure 2:

Kaplan-Meier curve of survival for patients treated with HIPEC or HIPEC + EPIC post-propensity matching. HIPEC, hyperthermic intraperitoneal chemotherapy; EPIC, early postoperative intraperitoneal chemotherapy.

Discussion

This data shows a significant median survival benefit of 34.3 (127.3 vs. 93.0, p=0.0213) months with the addition of EPIC to HIPEC in a matched propensity study in patients with peritoneal spread of low grade appendiceal tumours. This was also associated with a median increase of 1.5 days total length of hospital stay (23.5 vs. 25.0, p=0.028).

There exists limited literature on the comparison of HIPEC to HIPEC + EPIC for LAMN. Sparks et al. showed no significant difference in overall survival for appendiceal cancer between HIPEC and HIPEC + EPIC, but combined both high and low grade in analysis and was likely underpowered to detect a difference (n=13 and 17 respectively, p=0.645) [17]. For HAMN, Lam et al. analysed survival for both colorectal and HAMN showing no significant survival benefit, but trend to better survival with 50% survival at 3 years compared to 46% for HIPEC + EPIC to HIPEC alone (p=0.72) [18]. As the included appendiceal tumours were high grade and different from our cohort, as well as making up only 25.8% (n=24) patients in their analysis, these are not translatable to our cohort. Chua et al. showed both appendiceal and colorectal cancers not to have survival benefit with additional EPIC to HIPEC, only showing disease free survival benefit with additional EPIC in cancers of colorectal origin [19], once again of differing pathology to our cohort. For appendiceal neoplasms with PMP alone; Soucisse et al. analysed the use of HIPEC with and without EPIC using propensity matching, identifying no benefit of EPIC even for LAMN, but noted those who had EPIC were from earlier in their centres learning curve and had higher mean CC score [20]. Huang et al. showed additional EPIC to be associated with improved OS for high-grade appendiceal cancer with a HR of 0.45 (p=0.004), but excluded those of low-grade [12]. Fung et al. in their series of HIPEC with and without EPIC for both low and high grade appendiceal cancer with PMP showed no survival difference, but also excluded patients undergoing high tumour debulking [21]. An additional study using an older version of this dataset in LAMN showed a 30.5% five-year overall survival benefit (93.0% for HIPEC+ EPIC vs. 64.5% for HIPEC alone, p<0.001), with median survival not reached [11]. Our study with its propensity score matching extends on this analysis and provides further evidence into the potential benefits of EPIC in LAMN with PMPS

Our major morbidity rates of 40.1% in the HIPEC+EPIC group was not different from the HIPEC group, pre-propensity matching (32.7%, p=0.492). Concern for higher morbidity with EPIC is common in the literature as a reason it fell out of favour, with trials from Tan et al. showing higher morbidity (58% vs. 25%, p=0.048), Glehen et al. showing higher rates of fistulas with a risk ratio of 1.7, p=0.032 and Lam et al. finding additional EPIC as an independent predictor of morbidity, with morbidity rates of 43.2% with EPIC vs. 19.6% of HIPEC alone [18, 22, 23]. Our morbidity rates in the HIPEC only group are higher than the reported rates in other series, but given our mean HIPEC PCI of 24.6, likely mirrors the extent of resection, with previous analyses of morbidity showing PCI>26 to be an independent predictor [24]. In comparison, Tan et al. with a morbidity rate of 25% in their HIPEC group reported a median PCI of 15 and 9 in their HIPEC cohorts, with likely a lower overall mean PCI [22]. It has been noted that recent work comparing HIPEC with HIPEC+EPIC appears to not show increased morbidity compared to older papers, possibly indicating better postoperative care [10].

Table 1 demonstrated differences in the baseline unmatched data of CC score and median ASA, which was not seen in the propensity matched cases (Table 2). This is likely a reflection of the strength of the study design, with propensity matching allowing these presumed covariates to be corrected for. The noted difference in age with those receiving EPIC being on average 6.2 years younger pre-matching also trended to correction, reducing to 4.1 years post-propensity matching. Age was not shown to impact survival at any stage of matching with a non-significant Pearson correlation of 0.018 and −0.05 pre and post matching respectively (Table 3). Whilst still a significant difference (p=0.044), the lack of impact on survival indicates this difference does not appear clinically significant in our cohort. Analyses from Glehen et al. had previously shown age not to affect survival until 65 years of age [23]. Similarly, CC score was also significantly lower in the HIPEC+EPIC group at baseline with evidence of impact on survival (Pearson correlation 0.231, p<0.001) (Tables 1 and 2), but this difference disappeared post-propensity matching with the matched group indicating that differing CC clearance between the groups did not affect survival. This likely explained by the median CC score being 0 in the matched groups, with less variance in scoring to allow for detection of CC impacting survival, which is a known factor.

Choice of HIPEC agent differed between the HIPEC alone and HIPEC+EPIC groups, with those receiving EPIC more likely to receive mitomycin C as opposed to oxaliplatin or cisplatin in both pre and post-propensity matching (Tables 1 and 2). This impacted survival pre-matching in multivariate analysis but propensity-matched survival was not affected by this choice, possibly highlighting the impact of covariables not corrected for in the unmatched influencing this. A large multicentre trial comparing oxaliplatin and mitomycin C in appendiceal tumours has previously shown no survival difference [25].

Finally, our total length of stay differed significantly in our post-propensity matched group, with the HIPEC + EPIC group experiencing a median longer length of stay when compared to HIPEC alone, at 25.0 vs. 23.5 days, p=0.028. ICU length of stay was not different. It is noted there are concerns of increasing ICU resources required for delivery of EPIC, but this was not shown in our data [10]. Shorter cycles of EPIC (1–2 days) have previously been shown to have no difference in survival benefits when compared to long course (3–5 days) EPIC [26]. The overall survival benefits likely outweigh the short median increase in length of stay for additional EPIC to HIPEC.

Conclusions

In LAMN with PMP, the addition of EPIC to HIPEC with CRS significantly improves overall survival in propensity score matched cases and should be considered in select patients.