Does having a gastrectomy delay time to feeding and prolong hospital stay in patients undergoing cytoreductive surgery and hyperthermic intraperitoneal chemotherapy?

Abstract

Aim: Cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC) is routinely used to treat selected patients with peritoneal carcinomatosis, but can be associated with prolonged hospital stay, significant morbidity and mortality. Our objective was to assess whether patients undergoing gastrectomy as part of CRS/HIPEC were at increased risk of delayed feeding time and prolonged hospital stay.

Methods: Two hundred and fourteen consecutive patients with peritoneal carcinomatosis treated with CRS/HIPEC between 2001 and 2016 were stratified by whether CRS included gastrectomy (n = 19, 9%) and compared. Primary outcomes were time to full feeds and rate of serious morbidity (Clavien–Dindo grades III–V). Secondary outcomes were durations of ICU and hospital stays.

Results: Of 214 patients undergoing CRS/HIPEC, those undergoing gastrectomy (19, 8.9%) had increased time to full feeds (8 vs. 5 days, p < 0.01), and duration of ICU (2 vs. 1 days, p < 0.01) and total hospital stays (16 vs. 14 days, p = 0.013). There was no significant increase in serious complications, although increased risk of pneumonia was noted (21% vs. 4.1%, p = 0.011). Undergoing gastrectomy was not independently prognostic in multivariable analysis including high peritoneal tumour load (PCI >12), multiple CRS procedures (number >2) and operation duration (>480 min) in which operative duration remained independently prognostic (p < 0.01).

Conclusions: After surgery, early oral refeeding may be beneficial in the majority of patients undergoing CRS/HIPEC. However, patients found to have high peritoneal tumour load with extended surgery and those who underwent gastrectomy should be considered for early post-operative TPN due to the significant risk of delayed time to full feeds.

Keywords:

• Regional perfusion cancer chemotherapy
• cytoreduction surgical procedures
• gastrectomy
• enteral nutrition
• peritoneal neoplasms

Introduction

Cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) have achieved good long-term results in treating peritoneal carcinomatosis from colorectal [1], ovarian [2], appendiceal [3], mesothelioma [4] and primary peritoneal neoplasms. An increase in overall survival has been reported at the expense of considerable peri-operative morbidity and mortality although patients are able to return to baseline or improved quality of life by 6–12 months [5]. This high complication rate has been attributed to the steep learning curve associated with the procedure, although even in several high volume centres, the complication rate remains significant and is comparable to major gastrointestinal surgery [6].

Major morbidity in CRS/HIPEC varies widely 12 [7]–67.6 [8]%, with a median of 31% reported from an analysis of several studies in the literature [9]. Rates of peri-operative mortality vary from 0 [10] to 9 [11]%, with a median of 4% [9].

An analysis of gastrointestinal complications in patients treated with CRS/HIPEC reported on an overall gastrointestinal morbidity rate of 17% although the cohort of 147 patients included only four (3%) undergoing gastrectomy [12]. Chua et al. reviewed studies reported before 2008 and drew comparisons to similar rates of digestive morbidity seen with other major gastrointestinal interventions (e.g. the Whipple procedure) especially in high volume centres [6]. The most common gastrointestinal complication is digestive fistula or anastomotic leak with reported rates ranging from 3.9 to 34% [3,13–15]. Halkia et al. reported that out of 185 patients treated with CRS/HIPEC, 16 patients (8.6%) developed digestive fistulas with spontaneous closure with total parenteral nutrition (TPN) being achieved in 11 (68.7%) [15].

Given the relatively high rate of gastrointestinal complications in CRS/HIPEC and the limited number of patients undergoing gastrectomy included in these studies, we aimed to assess whether patients undergoing gastrectomy as part of CRS/HIPEC were at increased risk of perioperative complications, primarily delayed feeding time and prolonged hospital stay.

Methods

Patients

A prospectively maintained, institutional review board approved database of all patients who underwent CRS/HIPEC for peritoneal-based malignancies at the National Cancer Centre Singapore from April 2001 through to March 2016, was retrospectively reviewed. Demographics including age, gender, race and tumour type were included in the database and are reported.

Patients considered for CRS/HIPEC had to be of Eastern Cooperative Group (ECOG) performance status 0 or 1, with no distant metastases. All patients were recommended for CRS/HIPEC after evaluation in a multidisciplinary tumour board. The extent of disease in the abdomen and pelvis was examined on computed tomography (CT) scan and the absence of extra-abdominal disease either via thorax CT or positron emission tomography (PET)-CT scan was determined. Values of albumin taken to represent mild, moderate and severe malnourishment were <35, <26 and <21 g/L, respectively.

CRS/HIPEC proceeded according to previously published techniques. Burden of disease at time of surgery was calculated according to the peritoneal carcinomatosis index (PCI) [16]. Complete cytoreduction was attempted whenever possible. Chemotherapy was infused via a hyperthermia pump (Belmont) into a closed abdomen at a target temperature of 41–42 °C for 60 min. The chemotherapeutic agent used was determined by the primary surgeon and medical oncologist on the basis of primary malignancy.

Primary outcome was time to full feeds and peri-operative complication rates. Secondary outcomes were durations of intensive care unit (ICU) and hospital stays. As complications were primarily postoperative they were categorised according to the Clavien–Dindo postoperative complication classification, with major complications defined as Clavien III and IV rather than NCI-CTCAE classification which is more widely used to evaluate the toxicity of chemotherapy and radiotherapy.

For purposes of comparison, CRS/HIPECs were grouped based on whether or not they included a gastrectomy and demographic and peri-operative features were compared between groups.

Data were analysed at the procedural level in order to increase the generalisability of results to include patients who underwent multiple CRS/HIPEC. In these instances, listed patient characteristics are representative of the patient’s state at the time of each included operation.

Statistics

Continuous variables were expressed as mean ±1 standard deviation (normally distributed data) and categorical data as proportions throughout the article. Clinical variables and surgical outcomes were compared across groups as follows. The Chi-square or Fisher's exact test (if a group contained five or less samples) for comparison between categorical variables reported as number and percentage in each group. Independent sample t-test or ANOVA with Tukey’s post hoc testing (when the comparison included more than one degree of freedom) for continuous variables, reported as mean ±1 standard deviation or using the Mann–Whitney U test for ordinal variables reporting median and interquartile range (IQR). A multivariable logistic regression model was used to assess independence of gastrectomy and outcomes. p < 0.05 was considered statistically significant. All statistical analyses were performed using R 3.2.3 (R Foundation, Vienna, Austria) [17].

Results

Baseline characteristics of study population

From April 2001 to February 2016, a total of 214 CRS/HIPECs were performed on 201 patients. Baseline demographics of the cohort are summarised in Table 1. Twelve patients underwent CRS/HIPEC twice and one patient underwent CRS/HIPEC three times. The majority of patients had primary tumours that were ovarian (30%), colorectal (29%) or appendix (22%). All CRS procedures included HIPEC. One hundred and fourteen (53%) of the HIPEC regimen included mitomycin C, 92 (43%) used cisplatin, eight (4%) used oxaliplatin. Overall rates of incomplete cytoreduction (2%), serious morbidity (23%) and 30-day mortality (0%) were comparable to previously reported data. The median time to full feeds was five days. There was no significant difference in demographics or outcomes for patients undergoing gastrectomy for gastric primary cancer with limited peritoneal disease (n = 4, median PCI 13) compared to patients undergoing gastrectomy for colorectal or appendiceal cancer with extensive peritoneal disease (n = 16, median PCI 21.5).

Table 1. Patient demographics.

Category	Parameter	Mean ± SD	Range	Median
Patient characteristics	Age (years)	51 ± 12	10–76	53
	Gender	56 Male (26%)	158 Female (74%)
Intra-operative parameters	Duration (hours)	500 ± 170	185–980	500
	PCI score	13 ± 9.1	0–39	12
Surgical procedures	128 sub-diaphragmatic stripping (60%), 115 colectomy (54%), 59 small bowel resection (28%), 53 cholecystectomy (25%), 54 splenectomy (25%), 36 THBSO (17%), 19 gastrectomy (8.9%), nine bladder/ureteric resection (4.2%), 107 others (50%)
Histology	53 adenocarcinoma (25%), 35 mucinous (16%), 24 serous (11%), 103 other (48%)
Primary	65 ovarian (30%), 63 colorectal (29%), 48 appendix (22%), 15 primary peritoneal (7%), 10 mesothelioma (5%), 13 other (6%)
Short-term post-operative outcomes	Serious morbidity	50 (23%)
	30 day inpatient mortality	0 (0%)
	Incomplete cytoreduction (CC score 2)	4 (2%)
	ICU stay	2 ± 6	0–76	1
	Total hospital stay	19 ± 18	6–188	14
	Time to full feeds	7 ± 5	2–41	5

PCI: peritoneal carcinomatosis index; THBSO: total hysterectomy bilateral salpingo-oophorectomy; ICU: intensive care unit.

Operations were grouped based on whether or not a gastrectomy was performed. Of 19 patients underwent gastrectomy, 18 underwent subtotal gastrectomy and one underwent total gastrectomy. Baseline clinical features of the two groups are summarised in Table 2. Patients undergoing gastrectomy were more commonly males (42% vs. 23%, p < 0.01) or had appendiceal tumours (58% vs. 19%, p < 0.01) compared to patients who did not undergo gastrectomy. Comorbidities were similar between patients who underwent gastrectomy and patients who did not. Although patients undergoing gastrectomy had lower pre-operative albumin (36 vs. 39, p = 0.021), there was no significant difference in nutrition status (p = 0.079), or pre-operative BMI (p = 0.47).

Table 2. Demographics by gastrectomy status.

	Gastrectomy (n = 19)	No gastrectomy (n = 195)	p value
Age (years)	50 ± 12	52 ± 12	0.54
Female:male	8:11	150:45	<0.01
Race			0.051
Chinese	12 (63%)	152 (78%)	–
Others	3 (16%)	26 (13%)	–
Indian	1 (5.3%)	12 (6.2%)	–
Malay	3 (16%)	5 (2.6%)	–
Comorbidities
Hypertension	4 (21%)	46 (24%)	1
Diabetes	2 (11%)	19 (9.7%)	1
Ischaemic heart disease	0 (0%)	2 (1%)	1
COPD	0 (0%)	1 (0.51%)	1
Asthma	0 (0%)	5 (2.6%)	1
Other malignancy	0 (0%)	18 (9.2%)	0.34
ECOG			1
0	15 (94%)	179 (92%)	–
1	3 (6%)	15 (8%)	–
Pre-operative albumin (g/L)	36 ± 7.1	39 ± 5.5	0.021
Malnourished (albumin <35 g/L)	6 (32%)	27 (15%)	0.13
Malnourishment status			0.11
None	13 (68%)	151 (85%)	–
Mild	5 (26%)	20 (11%)	–
Moderate	0 (0%)	5 (3%)	–
Severe	1 (5%)	2 (1%)	–
Pre-operative BMI	25 ± 4.1	24 ± 4.8	0.47
Primary tumour			<0.01
Appendix	11 (58%)	37 (19%)
Colorectal	4 (21%)	59 (30%)
Ovarian	1 (5.3%)	64 (33%)
Primary peritoneal	0 (0%)	15 (7.7%)
Others	3 (16%)	10 (5.1%)
Mesothelioma	0 (0%)	10 (5.1%)

COPD: chronic obstructive pulmonary disease; ECOG: Eastern Cooperative Group; BMI: body mass index.

Operative procedures

Details of the CRS/HIPECs are summarised in Table 3. Patients undergoing gastrectomy tended to have more extensive disease as indicated by higher median PCI score (20 vs. 10, p < 0.01) and longer mean operation time (590 vs. 490 min, p = 0.018). Of the additional procedures performed, gastrectomy was associated with increased rates of splenectomy (53% vs. 23%, p < 0.01) and cholecystectomy (47% vs. 23%, p < 0.035). The more frequent use of mitomycin C in patients undergoing gastrectomy is consistent with more patients with underlying gastrointestinal primary histologies (appendiceal and colorectal) being more likely to undergo a gastrectomy compared to patients with ovarian cancer as a primary disease.

Table 3. Comparison of details of CRS/HIPEC by gastrectomy status.

	Gastrectomy (n = 19)	No gastrectomy (n = 195)	p value
PCI score (median)	20 [14–28]	10 [5–18]	<0.01
PCI score ≥13	12 (86%)	72 (44%)	<0.01
Incomplete cytoreduction (CC ≥2)	1 (6.7%)	3 (1.6%)	0.26
Duration (minute, mean)	590 ± 160	490 ± 170	0.018
Number of procedures (median)	4 [3–6]	2 [2–3]	<0.01
Additional procedures
Subdiaphragmatic stripping	12 (63%)	116 (59%)	0.95
Colectomy	13 (68%)	102 (52%)	0.27
Small bowel resection	4 (21%)	55 (28%)	0.69
Splenectomy	10 (53%)	44 (23%)	<0.01
THBSO	0 (0%)	36 (18%)	0.083
Cholecystectomy	9 (47%)	44 (23%)	0.035
Bladder/ureter resection	1 (5.3%)	8 (4.1%)	1
HIPEC regime			0.033
Mitomycin C	15 (79%)	98 (51%)
Platinum agent	4 (21%)	96 (49%)

CC: completeness of cytoreduction score; CRS: cytoreductive surgery; HIPEC: hyperthermic intra-peritoneal chemotherapy; PCI: peritoneal carcinomatosis index; THBSO: total hysterectomy bilateral salpingo-oophorectomy.

Outcomes

Peri-operative outcomes and complications are summarised in Table 4. Gastrectomy was associated with higher median estimated blood loss (1500 ml vs. 1000 ml, p = 0.041) and post-operative transfusion requirements (2 vs. 1 unit, p < 0.01). There was an increase in median ICU and total hospital stays (2 vs. 1 day and 16 vs. 14 days, respectively), as well as an increase in time to full feeds (8 vs. 5 days, p < 0.01). Although gastrectomy was not associated with a significant overall increased risk of serious complications (32% vs. 23%, p = 0.55), there was an increased risk of pneumonia (21% vs. 4.1%, p = 0.014), with no significant change in other complications.

Table 4. Comparison of peri-operative outcomes by gastrectomy status.

	Gastrectomy (n = 19)	No gastrectomy (n = 195)	p value
Estimated blood loss (ml)	1500 [850–4200]	1000 [550–2000]	0.041
Intra-operative blood transfusion (units)	2 [0.5–8]	2 [1–4]	0.77
Post-operative blood transfusion (units)	2 [1–5]	1 [0–2]	<0.01
ICU stay (days)	2 [1–4.5]	1 [0–2]	<0.01
Time to full feeds (days)	8 [6–10]	5 [4–7]	<0.01
Total hospital stay (days)	16 [14–38]	14 [11–19]	0.013
Minor complications (grade I–II)	9 (47%)	85 (44%)	0.94
Serious complications (grade III–V)	6 (32%)	44 (23%)	0.55
30-day inpatient mortality	0 (0%)	0 (0%)	1
Complications
Respiratory	6 (32%)	32 (16%)	0.18
Pneumonia	4 (21%)	8 (4.1%)	0.014
Intra-abdominal collection	1 (5.3%)	20 (10%)	0.7
Enterocutaneous fistula	0 (0%)	4 (2.1%)	1
Enterovesicular fistula	0 (0%)	1 (0.51%)	1
Anastomotic leak	2 (11%)	4 (2.1%)	0.09
Acute renal impairment requiring dialysis	0 (0%)	5 (2.6%)	1
Acute renal impairment not requiring dialysis	1 (5.3%)	16 (8.2%)	1
Bleeding	1 (5.3%)	3 (1.5%)	0.37
Scrotal ulcer requiring debridement	0 (0%)	1 (0.51%)	1
Anastomotic stenosis requiring bypass surgery	1 (5.3%)	0 (0%)	0.089
Guillain–Barré	1 (5.3%)	0 (0%)	0.089
Chemical peritonitis	0 (0%)	1 (0.51%)	1
Wound infection	1 (5.3%)	13 (6.7%)	1
Ileus	1 (5.3%)	8 (4.1%)	0.57

ICU: intensive care unit.

Multivariate analysis

Multivariate analysis of outcomes associated with gastrectomy on univariate analysis is summarised in Table 5. Outcomes were compared to whether patients underwent gastrectomy, PCI score ≥13, duration of operation ≥480 min and number of cytoreductive procedures ≥3. Gastrectomy was not independently associated with any of the outcomes assessed, although there was a trend to association with time to full feeds (median 8 vs. 5 days, p = 0.07). The strongest association was identified for duration of operation ≥480 min, which was independently associated with ICU stay (median 2 vs. 0 days, p < 0.01) and time to full feeds (median 6 vs. 5 days, p < 0.01).

Table 5. Multivariate analysis of outcomes associated with gastrectomy.

	No	Yes	Univariate p value	Multivariate p value
ICU stay (days)
Gastrectomy	1 [0–2]	2 [1–4.5]	<0.01	0.27
PCI ≥13	0 [0–1]	2 [1–2.2]	<0.01	0.96
Duration ≥480 min	0 [0–1]	2 [1–3]	<0.01	<0.01
Procedure number ≥3	0 [0–1]	2 [1–3]	<0.01	0.01
Total hospital stay (days)
Gastrectomy	14 [11–19]	16 [14–38]	0.013	0.91
PCI ≥13	12 [10–14]	16 [13–25]	<0.01	0.64
Duration ≥480 min	12 [10–15]	16 [13–27]	<0.01	0.95
Procedure number ≥3	12 [10–15]	16 [12–27]	<0.01	0.06
Time to full feeds (days)
Gastrectomy	5 [4–7]	8 [6–10]	<0.01	0.07
PCI ≥13	5 [4–6]	6 [5–8.8]	<0.01	0.04
Duration ≥480 min	5 [4–6]	6 [5–8.8]	<0.01	<0.01
Procedure number ≥3	5 [4–6]	6 [5–9]	<0.01	0.39
Pneumonia (days)
Gastrectomy	8 (4.1%)	4 (21%)	0.014	0.20
PCI ≥13	1 (1.1%)	6 (7.1%)	0.052	0.18
Duration ≥480 min	0 (0%)	12 (11%)	<0.01	0.81
Procedure number ≥3	0 (0%)	12 (11%)	<0.01	0.81

ICU: intensive care unit; PCI: peritoneal carcinomatosis index.

Discussion

In our study, patients who underwent a gastrectomy as part of CRS/HIPEC tended to have more advanced disease, with higher tumour volume as evidenced by a higher PCI score (20 vs. 10). Although the need for gastrectomy may represent a different disease burden between colorectal or appendiceal cancers with more extensive peritoneal disease than in gastric primary cancers (PCI 21.5 vs. 13), there was no significant difference in outcomes between these two groups. Higher PCI score naturally correlated with a longer duration of surgery (590 vs. 490 min), incorporation of a greater number of CRS procedures (4 vs. 2), as well as longer ICU (2 vs. 1 days) and hospitalisation stays (16 vs. 14 days). An increase in the time to full feeds (8 vs. 5 days) was also noted, which may be contributed to the presence of a gastrectomy as well as a likely longer recovery time and prolonged ileus in those who had undergone more extensive surgery for their more advanced disease.

There is limited evidence in the literature on the basis of which to make nutritional recommendation in CRS/HIPEC, although pre-operative malnourishment has been associated with a longer length of stay (15.0, 15.2 and 27.8 days for well nourished, mild-moderately malnourished and severely malnourished, respectively), without an increased risk of complications [18].

Further information on feeding regimes in CRS/HIPEC patients is limited, but some comparisons can be made to peri-operative care in patients undergoing gastrointestinal resections whereby no definite consensus has been reached on the role of early vs. traditional post-operative feeding as well as that of peri-operative TPN. A meta-analysis of early vs. traditional post-operative feeding in patients undergoing gastrointestinal resections reported on a total of 1240 patients [19]. Early post-operative feeding was associated with a reduction in odds of post-operative complications (OR 0.55) without effect on anastomotic dehiscence or reduced length of stay. However, only eight patients underwent a gastrectomy in this series. A similar meta-analysis of 454 patients reported on early oral feeding (EOF) after gastrectomy for gastric cancer [20]. No difference was reported in post-operative complications, but a reduced length of hospital stay was reported in the EOF group (reduced by 2.36 days).

Further studies have failed to establish similar results in patients undergoing gastrectomy. Sierzega et al. reported on 353 patients undergoing total gastrectomy for cancer, in which 135 (52%) had EOF [21]. There was no increase in the risk of anastomotic failure (OR 0.924, p = 0.709), and early feeding was associated with fewer surgical (15 vs. 24%) and general (8 vs. 23%) complications. However, subsequent multivariate regression models failed to confirm a significant correlation between timing of oral meals and post-operative morbidity.

In patients with malnourishment undergoing gastrectomy, peri-operative use of TPN has shown benefits for reducing morbidity and length of hospital stay, particularly when used peri-operatively. A study by Wu et al. reported that in 118 malnourished patients undergoing total (n = 40) or subtotal (n = 78) gastrectomy, patients who did not receive TPN had higher morbidity (67 vs. 16% and 44% vs. 22% for subtotal and total gastrectomy, respectively) as well as a longer hospital stay (35 vs. 21 days) [22]. In addition, they noted that the morbidity rate and length of hospital stay were higher in patients who received post-operative TPN only compared to patients receiving peri-operative TPN.

Our study demonstrates that the need for a gastrectomy belies more significant peritoneal disease in patients undergoing CRS/HIPEC. Patients who undergo extensive abdominal surgeries with gastrointestinal resections are often already at risk for delayed feeding, and the addition of a gastrectomy likely contributes to a further prolongation of the time to full feeds, although the study is limited by the low number of patients undergoing gastrectomy and is unable to definitively demonstrate this. It is important that nutritional re-establishment is included in patients recuperating from these surgeries and if delayed feeding can be predicted either in the pre-operative or early post-operative period, then a protocol that incorporates either the introduction of TPN from pre- or early post-operative period or one that examines the role of EOF should be instituted. With the results of this study and in light of the evidence provided by the available literature, our intention is to consider pre-operative nutritional supplementation if extensive disease or need for gastrectomy can be predicted from radiological imaging, and/or a deliberate attempt of early oral refeeding in patients undergoing CRS/HIPEC, with initiation of TPN on post-operative days 1 or 2 in those with a high PCI and who underwent a gastrectomy as part of their surgical procedure.

Conclusions

The need for gastrectomy in CRS/HIPEC is associated with more extensive peritoneal disease and longer operation time resulting in delayed feeding and prolonged hospital stay. Whilst further work is required to optimise peri-operative nutrition in CRS/HIPEC, the identification of pre-operative factors such as the extent of disease and likelihood of requirement for gastrectomy as part of CRS/HIPEC may aid in predicting which patients are candidates for pre-operative nutritional supplementation or EOF as well as patients at risk of significantly delayed time to full feeds for whom peri-operative or early post-operative TPN should be considered. For malnourished patients at significant risk of delayed time to full feeds, peri-operative TPN can be considered given the potential for lower morbidity rates compared to malnourished patients receiving post-operative TPN only.

Disclosure statement

All authors declare no conflict of interest or receive any funding for research.