World Journal of Surgery

, Volume 34, Issue 1, pp 70–78

A Critical Evaluation of Risk Factors for Complications After Cytoreductive Surgery and Perioperative Intraperitoneal Chemotherapy for Colorectal Peritoneal Carcinomatosis

Authors

  • Akshat Saxena
    • Department of Surgery, St George HospitalUniversity of New South Wales
  • Tristan D. Yan
    • Department of Surgery, St George HospitalUniversity of New South Wales
    • Department of Surgery, St George HospitalUniversity of New South Wales
Article

DOI: 10.1007/s00268-009-0206-0

Cite this article as:
Saxena, A., Yan, T.D. & Morris, D.L. World J Surg (2010) 34: 70. doi:10.1007/s00268-009-0206-0

Abstract

Background

Cytoreductive surgery (CRS) combined with perioperative intraperitoneal chemotherapy (PIC) has demonstrated improved survival in selected patients with colorectal peritoneal carcinomatosis (CRPC). This treatment modality is associated with relatively high rates of perioperative morbidity and mortality. This study evaluated the clinical and treatment-related risk factors for perioperative morbidity and mortality in patients with CRPC who underwent CRS and PIC.

Methods

Sixty-three consecutive patients who underwent CRS and PIC for CRPC were evaluated. Adverse events were rated from grades I to V with increasing severity. Clinical and treatment-related risk factors for grades III and IV/V morbidity were determined.

Results

There were no perioperative deaths (0%). The grades III and IV morbidity rates were 14 and 17%, respectively. A peritoneal cancer index >12 (p = 0.019), transfusion >4 units (p = 0.028), number of peritonectomy procedures >3 (p = 0.013), left upper quadrant peritonectomy procedure (p < 0.001), and number of primary colonic anastomosis >1 (p = 0.004) were associated with grade IV morbidity on univariate analysis. Only left upper quadrant procedure was associated with grade IV morbidity on multivariate analysis (p = 0.002). Only number of primary colonic anastomosis >1 (p = 0.037) was associated with grade III morbidity on univariate analysis. This also was associated with grade III morbidity on multivariate analysis (p = 0.028).

Conclusions

CRS and PIC has an acceptable risk of perioperative morbidity in carefully selected patients with CRPC. Patients who require extensive surgery have the highest risk for a severe adverse event. Preoperative evaluation of patients is essential to improve perioperative outcome.

Introduction

Colorectal peritoneal carcinomatosis (CRPC) was previously regarded as a form of systemic metastasis and treated palliatively. However, recent research has indicated that isolated CRPC is the result of transcoelomic invasion by the primary cancer or intraperitoneal seeding during surgical manipulation. Therefore, it should be viewed as a local-regional extension of disease rather than another manifestation of systemic metastasis [13]. Cytoreductive surgery (CRS) and perioperative intraperitoneal chemotherapy (PIC) have been used by several institutions as a treatment option for patients with isolated CRPC. Several phase II studies have confirmed the efficacy of this treatment modality and demonstrated a median survival of 15–63 months with a 5-year survival of 11–51% in selected patients, provided there is complete removal of macroscopic disease [46]. Moreover, a single, randomized, controlled trial from the Netherlands Cancer Institute confirmed the superiority of CRS and hyperthermic intraperitoneal chemotherapy (HIPEC) compared with palliative surgery and systemic chemotherapy in patients with isolated CRPC [7].

However assessment of survival outcome must be viewed against the operative risk that this treatment modality presents. In particular, early series reported a relatively high rate of perioperative mortality ranging from 0–12% and morbidity ranging from 23–44% [4]. This has contributed, in part, to the limited acceptance of this treatment modality amongst the general oncology community. A few studies have identified potential predictive factors for moderate–high-grade morbidity in patients who have undergone CRS and PIC. These include extent of carcinomatosis, [811] extent of surgical dissection [12, 13], duration of surgery [9, 11, 14], stoma formation [9], blood loss [811], number of bowel anastomoses [11], and age [15]. However, nearly all studies have combined the morbidity or mortality results for peritoneal surface malignancies from multiple sites together. These data do not allow a critical assessment of CRS and PIC as a treatment regime for CRPC.

The purpose of this study was to provide a thorough assessment of the mortality and morbidity outcomes of patients with CRPC who underwent the combined treatment. The correlations of clinical and treatment-related risk factors with moderate–severe morbidity were evaluated to allow more knowledgeable patient selection. As a result improved perioperative outcome, reduced hospital stay, and improved treatment acceptance is expected.

Methods

Selection criteria

Sixty-three consecutive patients underwent CRS and PIC for CRPC between February 1996 and February 2009. Informed consent was obtained from all patients. Patients were deemed suitable for CRS and PIC through consensus of a multidisciplinary team consisting of treating surgeons, medical oncologists, and radiologists at weekly meetings. All patients had biopsy-confirmed diagnosis of colorectal peritoneal carcinomatosis characterized clinically through exploration on laparotomy. Preoperative investigations performed to aid disease assessment included complete history, physical examination, tumor markers, and contrast-enhanced abdominal, pelvic, and chest CT.

CRS and PIC was offered to patients who were younger than 80 years, with good performance status (World Health Performance Status ≤ 2), and adequate hematological, hepatic, cardiac, and liver function. Patients with extra-abdominal metastasis were excluded. Extent of previous surgery was determined according to the number of abdominopelvic regions dissected during previous surgery [16].

Preoperative management

Patients were admitted the day before surgery. On admission, 5,000 units of subcutaneous heparin were administered twice per day to all patients. Blood samples were collected for analysis of hematological and coagulation parameters. Mechanical bowel preparation was performed. Cefotaxime 1 g and metronidazole 500 mg were administered approximately 1 hour before surgery and continued throughout the surgery. Alternating-calf compression boots were wrapped around the calves of patients to help prevent deep venous thrombosis. The anesthesia risk was assessed by using the American Society of Anesthesiologists (ASA) classification [17].

Cytoreductive procedures

All cytoreductive procedures were performed by or under supervision of a single surgeon (DLM). A midline incision from the xiphisternum to the pubic symphysis was performed to allow exploration of the abdominal cavity. The volume and extent of the tumor deposits was prospectively recorded using the Peritoneal Cancer Index (PCI) proposed by Sugarbaker [16]. This assessment combines lesion size (LS) (LS 0: no macroscopic tumor; LS 1: tumor < 0.5 cm; LS 2: tumor 0.5–5 cm; and LS3: tumor > 5 cm) with tumor distribution (abdominopelvic region 0–12) to quantify the disease as a numerical score (PCI 0-39). Peritonectomy procedures were then performed according to Sugarbaker’s guidelines [18]. These included total anterior parietal peritonectomy, omentectomy ± splenectomy, right and left upper quadrant peritonectomy, pelvic peritonectomy, lesser omentectomy with ± cholecystectomy. Visceral resections were performed at anatomic sites where tumor deposits were infiltrating deeply into an organ rendering surface excision ineffectual. These included rectosigmoidectomy, right colectomy, total abdominal colectomy, hysterectomy, small bowel resection, and partial gastrectomy. The purpose of CRS was to minimize residual disease. All sites and volumes of residual disease after CRS were recorded prospectively using the completeness of cytoreduction (CC) score. A CC-0 indicated no visible evidence of disease. CC-1 indicated residual tumors ≤ 2.5 mm in diameter. CC-2 indicated residual tumors between 2.5 mm and 2.5 cm in diameter. CC-3 indicated residual tumors > 2.5 cm in diameter or a confluence of tumor nodules present at any site [16].

Perioperative intraperitoneal chemotherapy

After the cytoreduction was completed, but before intestinal anastomosis or repair of seromuscular tears, the abdomen and pelvic were instilled with hyperthermic intraperitoneal chemotherapy (HIPEC, 10–12 mg/m2 mitomycin C) in the operating room at approximately 42°C in 3 L of 1.5% dextrose peritoneal dialysis solution for 90 minutes. The coliseum technique was used. The intraperitoneal chemotherapy solution was manually distributed by surgical staff to facilitate maximal contact and penetration of the chemotherapy into residual cancer cells.

Early postoperative intraperitoneal chemotherapy (EPIC, 650–800 mg/m2 5-flourouracil per day) was given in 1 L of 1.5% dextrose peritoneal dialysis solution from postoperative days 1 to 5. This was either in the intensive care unit (ICU) or high-dependency unit (HDU). The intraperitoneal chemotherapy was allowed to dwell for 23 hours and was then removed during the course of 1 hour. When the abdomen was cleared of fluid as completely as possible, the next instillation was commenced. In the present clinical pathway, all patients were scheduled to receive EPIC. However, in patients with hemodynamic instability or those experiencing early postoperative complications, EPIC was withheld. The criteria for EPIC include absence of leakage of the intraperitoneal chemotherapy system, absence of major organ failure, and the ability of patients to tolerate an increased intraabdominal fluid volume and intraperitoneal pressure with adequate urine output.

Postoperative management

No antibiotics were used postoperatively except to treat a specific organism causing postoperative infection. Prophylaxis for deep venous thrombosis was given in all patients. Doppler ultrasound was ordered if there were suspicions of a deep venous thrombosis. If patients developed respiratory symptoms and signs suggesting a pulmonary embolus, a ventilation/perfusion scan or CT pulmonary angiogram was performed. Oral contrast-enhanced abdomen and pelvic CT was performed in patients who had a clinical suspicion of intraabdominal collection or abscess. Total parenteral nutrition was used in patients with prolonged gastroparesis or inadequate caloric intake. Pneumococcal polysaccharide vaccine was administered in patients who had a splenectomy.

Study methods

A prospective database was designed to specifically record and evaluate the perioperative outcomes of patients with CRPC. Adverse events were graded 0-V in accordance with the National Cancer Institute’s Common Toxicity Criteria. Mild complications required medical or no treatment for resolution (grade I/II). Moderate complications required interventional procedures for resolution, such as a CT- or ultrasound-guided percutaneous drainage (grade III). Severe complications required returning to the operating room or intensive care support (grade IV). Perioperative death was defined as patient’s death within 30 days of surgery or during the same hospital admission (grade V).

Patient, procedural, and postoperative data were transferred into a new database. For analysis, serological variables (e.g., hemoglobin count, platelet count) were dichotomized on the basis of our institutional cutoff. Other variables (e.g., operation length, transfusion, age, peritoneal cancer index) were dichotomized based on their mean values (> mean, ≤ mean). Twelve clinical and 22 treatment-related factors were analyzed for an association with grades III and IV/V perioperative morbidity, respectively. The χ2 analysis or Fisher’s exact test, when appropriate, was used for univariate comparisons. All variables associated with grades III and grades IV/V morbidity with a p value ≤ 0.1 in univariate analysis were examined consecutively with multivariate analysis using a forward stepwise binary logistic regression model. Significance was defined as p < 0.05. Statistical analysis was performed using SPSS software (Version 16.0; GmbH, Munich, Germany).

Results

Descriptive data

A total of 63 consecutive patients who underwent CRS and PIC for CRPC between February 1996 and February 2009 were evaluated. Thirty-two patients (51%) underwent CRS and PIC after June 2006. The mean age of the study cohort at time of surgery was 55 (standard deviation (SD) = 15) years and 39 patients (62%) were women. Six patients (10%) had well-differentiated, 39 patients (62%) had moderately differentiated, and 18 patients (29%) had poorly differentiated tumors. The ASA classification was 1 in 5 patients (8%), 2 in 21 patients (33%), 3 in 31 patients (49%), and 4 in 2 patients (3%). In 4 patients (6%), no ASA score was recorded. The mean PCI per patient was 12 (SD = 8). Forty-five patients (72%) received a minimum of one cycle of systemic chemotherapy in the 12 months before their cytoreductive procedure. Twenty-four patients (38%) received a minimum of one cycle of systemic chemotherapy in the 3 months before their cytoreductive procedure and 21 patients (33%) received a minimum of one cycle of systemic chemotherapy 3 to 12 months before their cytoreductive procedure.

The mean operative duration was 7 (SD = 3) hours. The mean blood transfusion requirement was 4 (SD = 5). The mean number of peritonectomy procedures performed was 3 (SD = 1.5). Colonic resection was performed in 48 patients (76%). The mean number of colonic anastomosis performed was 1 (SD = 1.5). Other procedures performed included hysterectomy/bilateral salpingo-oophorectomy (n = 10, 16%), partial hepatectomy (n = 10, 16%), and partial gastrectomy (n = 2, 3%). Optimal cytoreduction (CC-0/CC-1) was achieved in 60 (95%) patients. Cytoreduction was suboptimal (CC-2/CC-3) in three (5%) patients. Thirty-four of the 63 patients (54%) received both HIPEC and EPIC as perioperative intraperitoneal chemotherapy. In 12 patients (19%) only HIPEC was administered. In 17 patients (27%) only EPIC was administered. The mean intensive care unit (ICU), high-dependency unit (HDU), and overall hospital stay was 4 days (SD = 7), 5 days (SD = 7), and 28 (SD = 29) days, respectively.

Perioperative mortality

There were no perioperative deaths (grade V).

Perioperative morbidity

Fourteen patients (22%) experienced no adverse events. Twenty-nine patients (52%) sustained 44 mild (grade I/II) adverse events, which required medical or no treatment for resolution. In the descending order of frequency, these mild adverse events included pleural effusion (n = 16), small bowel obstruction (n = 7), febrile episodes (n = 7), fistula (n = 4), cardiac arrhythmia (n = 3), pneumothorax (n = 3), pneumonia (n = 2), chemotherapy-related renal impairment (n = 1), and pulmonary embolism (n = 1).

Nine patients (14%) experienced 16 moderate (grade III) adverse events that required interventional procedures for resolution. Two patients (3%) had three grade III events and three patients (5%) had two grade III events each. Six patients (8%) developed an intra-abdominal abscess, which required CT-guided percutaneous drainage. Three patients (5%) had a symptomatic pleural effusion requiring insertion of a chest drain. Two patients (3%) had a prolonged ileus requiring intravenous therapy. Two patients (3%) had a prolonged fistula, which required total parenteral nutrition, somatostatin, and percutaneous drainage. One patient (2%) had a urinary tract infection with elevated temperatures and leukocytes. One patient (2%) had a ureteric leak that required stent insertion. One patient (2%) had a rapid atrial fibrillation, which required intervention therapy. Figure 1 demonstrates the relative incidence of grade III events encountered.
https://static-content.springer.com/image/art%3A10.1007%2Fs00268-009-0206-0/MediaObjects/268_2009_206_Fig1_HTML.gif
Fig. 1

Relative incidence of 16 grade III events in nine patients. In some patients, there was more than one grade III adverse event

Eleven patients (17%) experienced 15 severe (grade IV) events that required return to the operating room or intensive care support. One patient (2%) had three grade IV events and two patients (3%) had two grade IV events each. Four patients (6%) developed gross intra-abdominal sepsis, which required surgical washout. Three patients (5%) had postoperative hemorrhage, which required urgent return to operating room. Two patients (3%) developed acute bowel perforation. Two patients (3%) developed pneumonia, causing respiratory distress, which required endotracheal intubation. Two patients (3%) required reoperation for an anastomotic leak. One patient (2%) required reoperation for a pancreatic leak. One patient (2%) required thoracotomy and decortication for a multiloculated empyema. Figure 2 demonstrates the relative incidence of grade IV events encountered.
https://static-content.springer.com/image/art%3A10.1007%2Fs00268-009-0206-0/MediaObjects/268_2009_206_Fig2_HTML.gif
Fig. 2

Relative incidence of 15 grade IV events in 11 patients. In some patients, there was more than one grade IV adverse event

Univariate analysis of risk factors associated with moderate and severe morbidity

Table 1 shows the influence of 12 clinical factors on grades III and IV morbidity. There were no clinical factors associated with grade III morbidity. Only a peritoneal cancer index > 12 (p = 0.019) was associated with grade IV morbidity.
Table 1

Correlation of clinical factors with grades III and IV morbidity

Variable

Grade III morbidity n (%)

p

Grade IV morbidity n (%)

p

Total (n = 63)

9 (14)

11 (17)

Age at time of surgery (yr)

0.709

1.000

 ≤55 (n = 24)

4 (17)

4 (17)

 >55 (n = 39)

5 (13)

7 (18)

Gender

0.709

0.735

 Female (n = 39)

5 (13)

6 (15)

 Male (n = 24)

4 (17)

5 (21)

Peritoneal cancer index

0.454

0.019

 ≤12 (n = 38)

5 (13)

3 (8)

 >12 (n = 25)

4 (16)

8 (32)

ASA classification

0.269

0.488

 ≤2 (n = 26)

2 (8)

3 (12)

 >2 (n = 33)

6 (18)

7 (21)

Treatment period

0.475

0.750

 Before June 2006 (n = 31)

3 (10)

6 (19)

 After June 2006 (n = 32)

6 (19)

5 (16)

Extent of prior surgery

0.092

0.249

 >5 abdominopelvic regions dissected (n = 14)

4 (29)

4 (29)

 ≤5 abdominopelvic regions dissected (n = 49)

5 (10)

7 (14)

Recent preoperative systemic chemotherapy

0.322

0.842

 Yes (last dose 3–12 months before surgery) (n = 21)

3 (14)

3 (14)

 Yes (last dose 0–3 months before surgery) (n = 24)

5 (21)

5 (21)

 No (n = 18)

1 (6)

3 (17)

Preoperative hemoglobin (g/L)

0.157

 

1.000

 ≥125 (n = 34)

3 (9)

6 (18)

 <125 (n = 29)

6 (21)

5 (17)

Preoperative platelet count (x109)

1.000

0.828

 >150 (n = 56)

8 (14)

9 (16)

 ≤150 (n = 7)

1 (14)

2 (29)

Preoperative INR

0.590

0.371

 >1.1 (n = 10)

2 (20)

3 (30)

 ≤1.1 (n = 50)

7 (14)

8 (16)

Preoperative albumin (g/L)

1.000

0.676

 >32 (n = 49)

8 (16)

8 (16)

 ≤32 (n = 12)

1 (8)

3 (20)

Tumor differentiation

0.676

1.000

 Well/ moderate (n = 45)

5 (11)

8 (18)

 Poor (n = 18)

3 (17)

3 (17)

Table 2 shows the predictive influence of 22 treatment-related factors on grade III and IV morbidity. One treatment-related factor was associated with grade III morbidity: number of primary colonic anastomosis > 1 (p = 0.037). Four treatment-related factors were associated with grade IV morbidity: number of peritonectomy procedures > 3 (0.013), left upper quadrant peritonectomy (p < 0.001), transfusion > 4 units (p = 0.028), and number of primary colonic anastomosis > 1 (p = 0.004).
Table 2

Correlation of treatment-related factors with grades III and IV morbidity

Variable

Grade III Morbidityn (%)

p

Grade IV morbidityn (%)

p

Total (n = 63)

9 (14)

11 (17)

Operation length (hr)

0.482

0.184

 >7 (n = 32)

5 (16)

8 (26)

 ≤7 (n = 31)

4 (13)

3 (9)

No. of peritonectomy procedures

1.000

0.013

 >3 (n = 19)

2 (11)

7 (37)

 ≤3 (n = 44)

7 (16)

4 (9)

Anterior peritonectomy

1.000

0.263

 Yes (n = 47)

7 (15)

10 (21)

 No (n = 16)

2 (13)

1 (6)

Left upper quadrant peritonectomy

1.000

<0.001

 Yes (n = 12)

1 (8)

7 (58)

 No (n = 51)

8 (16)

4 (8)

Right upper quadrant peritonectomy

0.340

0.113

 Yes (n = 15)

3 (20)

5 (33)

 No (n = 48)

6 (13)

6 (13)

Omentectomy ± splenectomy

0.454

 

0.150

 Yes (n = 44)

5 (11)

10 (23)

 No (n = 19)

4 (21)

1 (5)

Pelvic peritonectomy

0.469

1.000

 Yes (n = 33)

6 (18)

6 (18)

 No (n = 30)

3 (10)

5 (20)

Lesser omentectomy ± cholecystectomy

0.435

0.494

 Yes (n = 22)

4 (18)

5 (23)

 No (n = 41)

5 (12)

6 (15)

Completeness of cytoreduction score

1.000

0.443

 Complete (CC–0/CC–1) (n = 60)

9 (16)

11 (17)

 Incomplete (CC–2/CC–3) (n = 3)

0 (0)

1 (33)

Transfusion (units)

0.192

0.028

 > 4 (n = 20)

4 (20)

7 (35)

 ≤ 4 (n = 43)

5 (12)

4 (9)

Partial hepatectomy

0.631

0.676

 Yes (n = 10)

2 (20)

1 (10)

 No (n = 53)

7 (13)

10 (19)

Hysterectomy ± bilateral oophorectomy

1.000

0.676

 Yes (n = 10)

1 (10)

1 (10)

 No (n = 53)

8 (15)

10 (19)

Colonic resection

0.421

0.272

 Yes (n = 48)

8 (17)

10 (21)

 No (n = 15)

1 (7)

1 (7)

Colonic resection procedure

0.250

0.314

 Colectomy + primary colonic Anastomosis ± ostomy (n = 39)

5 (13)

9 (23)

 Colectomy + ostomy only (n = 9)

3 (33)

 

1 (11)

 

 No resection (n = 15)

1 (7)

 

1 (7)

 

Primary colonic anastomosis

1.000

0.181

 Yes (n = 39)

5 (13)

9 (23)

 No (n = 24)

4 (17)

2 (8)

No. of primary colonic anastomosis

0.037

0.004

 >1 (n = 16)

4 (25)

7 (44)

 ≤1 (n = 47)

5 (11)

4 (9)

Ostomy

0.051

1.000

 Yes (n = 22)

6 (27)

4 (18)

 No (n = 41)

3 (7)

7 (17)

Colostomy

0.210

1.000

 Yes (n = 15)

4 (27)

2 (13)

 No (n = 48)

5 (10)

9 (19)

Ileostomy

0.184

0.595

 Yes (n = 7)

2 (29)

2 (29)

 No (n = 56)

7 (13)

9 (16)

Perioperative intraperitoneal chemotherapy

0.429

0.272

 HIPEC + EPIC (n = 34)

6 (18)

4 (12)

 HIPEC only (n = 12)

2 (17)

4 (33)

 EPIC only (n = 17)

1 (6)

 

3 (18)

 

Multivariate analysis of risk factors associated with moderate and severe morbidity

In the multivariate analysis, one factor was found to be independently associated with grade IV/V morbidity: left upper quadrant peritonectomy (risk ratio = 11.280, 95% confidence interval (CI) = 2.51–50.698, p = 0.002). One factor was found to be independently associated with grade III morbidity: number of primary colonic anastomosis > 1 (risk ratio = 6.080, 95% CI = 1.213–30.473, p = 0.028).

Discussion

Isolated colorectal peritoneal carcinomatosis (CRPC) usually develops through transcoelomic invasion by the primary cancer or intraperitoneal seeding during surgical manipulation. Therefore, it should be considered a local-regional extension of disease rather than a manifestation of systemic metastasis [13]. In carefully selected patients with CRPC, the combined approach of CRS and PIC has demonstrated improved survival outcomes compared with historical controls. Phase II studies have reported a median survival of 15–63 months and 5-year survival of 11–51% in patients who had a complete cytoreduction [46]. The single, randomized, controlled trial from the Netherlands Cancer Institute reported a significantly superior median survival (22.3 vs. 12.6 months) in patients who underwent CRS and HIPEC rather than palliative surgery with systemic chemotherapy [7].

However, this approach must be evaluated in terms of its risks. In particular, CRS and PIC for CRPC has been associated with a relatively high rate of perioperative mortality and morbidity in the past. Early case series reported a mortality rate between 0% and 12% and morbidity rate between 23% and 44% [4]. Our results from 13 years of experience with this procedure show that whilst there was no perioperative mortality, 32% of patients experienced a minimum of one moderate–severe adverse event. This reinforces the need to identify patients at high risk for moderate–severe morbidity to optimize patient selection and improve perioperative outcome.

Our study showed no correlation between any clinical factor and grade III morbidity. The only clinical factor associated with grade IV morbidity was a peritoneal cancer index (PCI) > 12. The PCI calculation numerically quantifies tumor burden and uniformly correlates with survival. Previous studies have shown that patients with high PCI derive minimal survival benefit from CRS and PIC [5, 8, 1922]. Furthermore, because the PCI is a surrogate marker for the complexity of surgery, it can help to identify patients at high risk of severe morbidity. Glehen and colleagues reported in a large multi-institutional study with 501 patients that the risk of severe complication in CRPC patients who underwent CRS and PIC was significantly increased by a PCI > 12 [8]. Similarly, Verwaal’s analysis of 102 patients with CRPC treated with CRS and HIPEC demonstrated a clear correlation between tumor load and surgical complication [10]. Our results correlate with these findings and question the value of the combined approach in patients with high volume CRPC particularly given their limited survival benefit.

Univariate analysis showed that four treatment-related factors were associated with a severe adverse event: number of peritonectomy procedures > 3, transfusion > 4 units, left upper quadrant peritonectomy procedure, and number of primary colonic anastomosis > 1. An extensive cytoreduction has been widely correlated with severe morbidity. Stephens and colleagues reported 200 cases of peritoneal carcinomatosis treated with CRS and HIPEC using the Coliseum technique. They found through multivariate analysis that the number of procedures and resections was the only significant risk factor for major morbidity [12]. Kusamara and colleagues conducted a study on 209 consecutive treatments of peritoneal carcinomatosis with CRS and HIPEC using the closed abdomen technique. Again, multivariate analysis identified extensive cytoreduction as a predictor for severe morbidity [13].

However, the pattern of disease spread within the abdomen also is important. Extensive disease involvement in the left hemidiaphragm is an indication for a left upper quadrant peritonectomy procedure. This procedure was the only significant predictor of severe morbidity on multivariate analysis. Left upper quadrant procedure involves stripping tissue beneath the left hemidiaphragm, cytoreduction on the spleen, and dissection around the tail of the pancreas. This procedure is associated with severe morbidity because diaphragmatic stripping commonly results in respiratory complications, such as atelectasis, leading to pneumonia and pleural effusion. Dissection on the splenic hilum and around the tail of the pancreas may predispose to bleeding, intra-abdominal abscess, and pancreatic leak. In our cohort, three patients (5%) developed severe pulmonary complications, and one patient (2%) required reoperation for a pancreatic leak. In all four patients, a left upper quadrant peritonectomy procedure was performed. In our institution, chest drains are prophylactically inserted in patients with diaphragmatic stripping to help prevent pleural effusions. Nonetheless, the procedure of chest drain insertion is not without its own inherent risk.

Bowel resection is indicated when the tumor nodules grossly involve the serosal surface of the small bowel and the accompanying mesentery. Unfortunately intestinal anastomosis is a high-risk procedure because intraperitoneal chemotherapy inhibits wound healing and surgery removes the peritoneum, which is the primary defense barrier for infection. Especially in a primary colonic anastomosis, there is increased risk of intra-abdominal abscess, pancreatic leak, and sepsis. These three complications accounted for 39% of grade III/IV adverse events. As expected, our study identified an increased risk in patients in whom > 1 primary anastomosis was performed. However, the relationship of surgical procedures to subsequent morbidity must be interpreted with caution. It is possible that moderate–severe adverse events associated with procedures may be related to more extensive cytoreductive surgery and prolonged operation duration. Similarly, the increased risk of severe morbidity in patients with high blood transfusion (> 4 units) may reflect a more extensive cytoreduction. However, it also may reflect an independent association as transfusion impairs various functions of cellular immunity. These include antigen processing, macrophage activation, T-lymphocyte function, and neutrophil and monocyte cytocidal activity [23]. Many studies have linked transfusion to an increased incidence of postoperative bacterial infections and severe morbidity [24, 25]. Therefore, reducing blood loss and/or subsequent transfusion through blood conservation strategies may improve perioperative outcome.

Studies in CRS and PIC procedures have identified other predictive factors for severe morbidity, including operation length [9, 11, 14], age [15], number of visceral resections [11], stoma formation [9], dose of chemotherapeutic agent [13], and recurrent cancer [10]. Most of these studies combined the perioperative outcomes of patients with peritoneal malignancies from various sites together. Others focused on survival as the primary end point. Given the importance of selecting patients with CRPC for CRS and PIC, a critical assessment of morbidity and mortality outcomes in this group of patients is necessary. Our study is one of the largest series of perioperative outcomes in patients who have undergone the combined treatment for this single diagnosis. Nevertheless, the number of patients is limited. This may have contributed to the fact that our study did not show a significant decrease in the rate of postoperative complications over time. The “learning curve” in the context of CRS and PIC has been demonstrated by two studies, both of which combined the perioperative outcomes of patients with peritoneal malignancies from various sites together [26, 27].

Another issue is that an effective comparison of morbidity outcomes between different institutions requires a consistency in treatment practice, postoperative management, and reporting of adverse events in a large number of patients. This has not been achieved. For example, there was no instance of grade III/IV hematological toxicity in our cohort of 63 patients. Other institutions report a rate of severe hematological toxicity between 2.5% and 19% and as much as 48% [28]. This can be explained in part by the low dose of mitomycin C (10–12.5 mg/m2) used at our institution. However, differences in postoperative management and reporting of adverse events also may be responsible.

To optimize patient selection and improve surgical planning before operation, a model that can predict the likelihood of severe events in patients would be useful. Several models for surgical patients are already being used. The Physiological and Operative Severity Score for enUmeration of Mortality and morbidity (POSSUM) uses 12 physiological and 6 operative variables to provide a calculated risk of morbidity and death [29]. Its efficacy in surgery involving removal of primary colorectal tumor has been confirmed by several large studies (> 5,000 patients) [30].

Conclusions

This critical assessment identified five clinical- and treatment-related variables associated with grade III/ IV morbidity on univariate analysis and two on multivariate analysis. It showed that patients who require extensive surgical dissection, in particular left upper quadrant peritonectomy and multiple bowel anastomoses, had the highest risk for moderate–severe morbidity. Whilst our study has utility in identification of high-risk patients, we believe that standardized management of patient and reporting of adverse events across different institutions would be most beneficial for patient selection. To achieve this collaboration between different peritonectomy centers and perhaps a multi-institutional prospective study with defined protocols is necessary.

Copyright information

© Société Internationale de Chirurgie 2009