Influence of Postoperative Complications on Relapse-Free Survival in Gastrointestinal Malignancies

Abstract

Objective:

A variety of preoperative variables—such as perforation prior to surgery, poor nutritional status, and comorbid conditions such as diabetes—are already known to shorten relapse-free survival in patients with gastrointestinal malignancies. However, the significance of postoperative events in gastrointestinal malignancies is still debated and has not been studied in the population of southern India.

Methods:

A retrospective study was conducted at Kidwai Memorial Institute of Oncology, Bangalore, India, from September 2004 to 2006. Patients from a single surgical unit who had undergone surgery with curative intent for gastrointestinal malignancies were evaluated (to maintain uniformity, patients who had undergone palliative surgery were not included in the analysis). We assumed anastomotic leak, delayed wound healing, and postoperative weight loss > 10% as risk factors predictive of poor disease-free survival. These factors were evaluated in all patients, and risk for development of relapse was calculated.

Results:

A total of 236 patients were evaluated. Baseline parameters were similar in both groups. Compared with patients who developed no postoperative complications, we found that the risk of relapse is 9.8 times greater in patients having anastomotic leak, 8.2 times greater in those with delayed recovery, and 2.3 times greater in those having excessive weight loss. The risk was uniform in all types of gastrointestinal malignancies.

Conclusion:

The results suggest that anastomotic leak, delayed wound healing, and postoperative weight loss in patients with gastrointestinal malignancies confer poor disease-free survival. The presence of these complications warrants closer follow-up and management as appropriate.

A variety of preoperative factors, such as perforation prior to surgery, poor nutritional status, and comorbid conditions such as diabetes, are known to be associated with poor relapse-free survival in patients with gastrointestinal malignancies. The mechanism underlying preoperative perforation is assumed to be migration of cancer cells into the peritoneum.¹ Similarly, multiple factors can contribute to shorter disease-free survival in diabetic patients, including the following complications:

• Insulin resistance with hyperinsulinemia and aberrant growth hormone–stimulating growth factor receptors

• Hyperglycemia leading to abnormal growth regulation

• Immune tolerance to malignant cells leading to early relapses²

In addition, poor nutritional status leading to treatment alterations and compromised delivery of postoperative adjuvant modalities, such as chemotherapy and radiotherapy, can be expected to exert an adverse effect on disease-free survival. The significance of postoperative events in gastrointestinal malignancies, however, is still debated and has not been studied in the population of southern India.

Although some postoperative complications, such as anastomotic leakage, have been identified as risk factors for early relapse, data are not uniform in all studies.^3–5 In one of the larger studies conducted, the statistical significance of anastomotic leak as predictive of overall survival was maintained even after adjusting for age, gender, emergency resection, site, size, stage, grade, and venous invasion (hazard ratio [HR] 1.6, 95% confidence interval [CI] 1.2–2.0).³ It has been suggested that increased local recurrences might be due, in part, to the presence of viable cancer cells at the time of surgery,³ and tumor spread and distant metastasis could be enhanced by an inflammatory response to anastomotic breakdown.^4,6 These popular theories, however, have not been formally demonstrated. Poor general patient condition, in fact, might contribute to increased risks of leak and also serve as an indirect marker of advanced disease, which suggests that leakage might actually be a causative aspect of relapse rather than a prognostic factor.¹

During the course of regular clinical practice, an author of this report identified two other postoperative factors that seemed to portend risk of early relapse—excessive weight loss and delayed recovery—that have not yet been studied. Accordingly, the authors conducted a retrospective study to investigate anastomotic leakage, excessive weight loss, and delayed recovery as potential postoperative risk factors for poor relapse-free survival in patients with gastrointestinal malignancies.

PATIENTS AND METHODS

This retrospective study was conducted at Kidwai Memorial Institute of Oncology, Bangalore, India, from September 2004 to 2006. Patients from a single surgical unit who had undergone surgery with curative intent for gastrointestinal malignancies were evaluated. To maintain uniformity, patients who had undergone palliative surgery were not included in the analysis.

For the purposes of this study, we hypothesized that the presence of anastomotic leak, delayed wound healing, and significant weight loss (> 10%) were risk factors predictive of relapse and poor disease-free survival. These factors were evaluated in all patients and risk of relapse was calculated. Final stage, presence of lymphovascular invasion, and type of cancer (anatomic site) were considered possible confounders. Cases with established prognostic parameters, such as advanced age, poor performance status, etc, that might confound the results were excluded from analysis. Univariate analyses to estimate the ability of each parameter to predict relapse (odds ratio [OR]) were performed using McNemar’s test. Multivariate analyses were performed using Cox’s regression analysis. All statistical analyses were performed using MedCalc^® version 10.1.0.0 for Windows (MedCalc Software, Broekstraat, Mariakerke, Belgium).

RESULTS

A total of 236 patients who underwent curative surgery for various gastrointestinal malignancies were evaluated. Baseline patient characteristics are shown in Table 1. Most patients underwent definitive surgery for their disease — surgical procedures and complications are summarized in Table 2. We found that type and duration of surgery did not vary significantly between patients. Similarly, no statistically significant differences in baseline characteristics existed between patients who developed complications and those who did not (Table 3).

Table 1.

Baseline patient characteristics

Characteristics	With relapse (n = 86)	Without relapse (n = 150)
Age in years (mean ± SD)	34.2 ± 102.6	36.9 ± 12.8
Sex (male:female)	1.8:1	1.6:1
Performance status (ECOG)
0	28	50
1	30	55
2	28	45
Primary diagnosis
Pancreas	12	12
Esophagus	16	18
Stomach	22	36
Colorectal	44	84
Stage at the time of surgery (AJCC group staging)
Stage I	22	30
Stage II	28	48
Stage III	36	72
Final stage (pathologic)
Stage I	18	25
Stage II	30	55
Stage III	38	80
Lymphovascular invasion	50%	33%
Grade (low:high)	2:3	3:4
Disease-free survival	20.2 months	Not reached

Abbreviations: SD = standard deviation; ECOG = Eastern Cooperative Oncology Group; AJCC = American Joint Committee on Cancer

Table 2.

Type of surgery and complication rates

	Pancreas (n = 24)	Colorectal (n = 128)	Stomach (n = 58)	Esophagus (n = 34)
Type of surgery	Whipple (24)	Right hemicolectomy (40)	D1 dissection (28)	Esophagectomy (17)
		Left hemicolectomy (38)	D2 dissection (15)	Level I dissection (10)
		Anterior resection (24)	Gastrectomy unclassified (15)	Level II dissection (5)
		Abdominoperineal resection (26)		Level III dissection (2)
Complications
Anastomotic leak	12	22	15	19
Wound infection	8	40	18	14
Weight loss	6	34	21	22
Mean time for recovery	20 days	14 days	16 days	18 days

Table 3.

Differences in characteristics of patients who had at least one postoperative complication vs. those who had no complications

Characteristics	With any complications (n=162)	Without complications (n=74)
Age in years (mean ± SD)	39.2± 9.2	34.6± 11.4
Sex (male:female)	1.5:1	1.8:1
Performances status (ECOG)
0	32	21
1	63	33
2	67	20
Primary diagnosis
Pancreases	20	4
Esophagus	22	12
Stomach	41	17
Colorectal	87	41
Stage at the time of surgery (AJCC group staging)
Stage I	42	10
Stage II	48	28
Stage III	72	36
Pathologic TNM Staging
T1/T2/T3/T4 (%)	15/25/35/25	25/35/25/15
N1/N2/N3 (%)	30/35/35	40/35/25
M0/M1 (%)	100/0	100/0
Lymphovascular invasion	56%	35%
Grade (low:high)	1:3.5	2:5
Disease-free survival	24.2 months	36.2
Percentage of patients receiving adjuvant therapy
Pancreas	75%	75%
Colorectal	88%	80%
Stomach	82%	78%
Esophagus	72%	76%

Abbreviations: SD = standard deviation; ECOG = Eastern Cooperative Oncology Group; AJCC = American Joint Committee on Cancer; TNM = tumor, node, metastasis

Upon final analysis, we found that, compared to patients who developed no postoperative complications, the risk of relapse is 9.8, 8.2, and 2.3 greater among patients having anastomotic leak, delayed recovery, or weight loss, respectively (Tables 4 and 5). Risk was uniform among all types of gastrointestinal malignancies. In the postoperative period, patients received adjuvant therapy in accordance with National Comprehensive Cancer Network (NCCN) guidelines (Table 6). The number of patients receiving chemotherapy did not vary significantly between the groups of those who developed complications vs. those who did not, as well as those who had relapsed or not. Thus, further analysis of the effect of adjuvant therapy was not performed.

Table 4.

Risk factors for recurrence

Risk factors	Odds ratio	95% Confidence interval	P value
Anastomotic leak (present/absent)	9.8	4.2–19.8	.01
Pancreas	2.8	0.7–9.2	NS
Esophagus	1.2	0.6–4.5	NS
Stomach	1	0.9–3.6	NS
Colorectal	1.1	0.8–8.6	NS
Final stage	1.3	0.7–6.8	NS
Lymphovascular invasion	1.1	0.9–2.6	NS

Abbreviation: NS = not significant

Table 5.

Multivariate analysis for final risk factors

Risk factors	Odds ratio	95% Confidence interval	P value	Odds ratio multivariate	95% Confidence interval	P value
Anastomotic leak	9.8	6.2–14.8	.004	3.8	2.1–10.1	.01
Postoperative weight loss	8.2	3.7–14.2	.03	2.3	1.9–9.6	.05
Delayed recovery	2.3	2.8–8.5	.05	1.1	0.9–4.5	NS

Abbreviation: NS = not significant

Table 6.

Adjuvant therapy received

	Pancreas (n = 24)	Colorectal (n = 128)	Stomach (n = 58)	Esophagus (n = 34)
Adjuvant CT alone	8	46	6	2
Adjuvant RT alone	2	0	7	18
Combined CT + RT	8	50	29	6
No adjuvant therapy	6	22	16	8

Abbreviations: CT = chemotherapy; RT = radiotherapy

DISCUSSION

Compared with patients who did not develop complications, the study results suggested that risk of relapse is 9.8 times greater in patients having anastomotic leak, 8.2 times greater in those with delayed recovery, and 2.3 times greater in those having weight loss.

Anastomotic Leak

The results of the present study are consistent with those reported by Walker et al.³ Of particular interest is the predictability of anastomotic leakage for poorer disease-free survival among all gastrointestinal malignancies (eg, pancreas, stomach, etc). Patients with leakage are at a 9.8 times greater risk for recurrence vs. those who did not. Similarly, the median progression-free survival (20.4 months) for those having anastomotic leakage is significantly lower (the end point had not been reached after 28 months’ median follow-up).

Weight Loss

The concept of weight loss as a negative prognostic factor is based on retrospective observation, where patients with significant weight loss occurring during the postoperative period experienced poorer overall outcome vs. patients who did not lose significant amounts of weight. We then hypothesized that circulating levels of tumor necrosis factor alpha and transforming growth factor, as well as interleukin-11, which are known to cause cancer cachexia, might be overexpressed in the immediate postoperative period if systemic spillage occurs during surgery. Thus, transient weight loss greater than expected (due to poor nutrition) might be a marker for relapse. Results showed that even after multivariate analysis, it maintained the statistical significance to predict poor outcome. As evaluated, however, the increase in levels of these markers was too small and the sample size was inadequate for further comment on this hypothesis.

Delayed Recovery

Initially, in univariate analysis, we found delayed recovery to predict poor outcome, but it lost its statistical significance in multivariate analysis. We assume, however, that delayed recovery, which is usually secondary to anastomotic leak, might serve as an apparent indicator of poor outcome.

Patients who developed any of the complications of interest had a significantly worse relapse-free interval (OR 2.8 favoring those who had no complications, P = .04) despite adjuvant therapy being given equally in both groups. These results indicate that adjuvant therapy did not influence disease outcome in the presence of the above-mentioned complications. Although it might be premature to recommend aggressive therapy in patients having these complications, it is well worth addressing this question in a prospective manner for formulating further guidelines.

Other Potential Factors

Patients with pancreatic cancer in our series were at greater risk of recurrence. Due to the relatively small numbers, however, it did not reach statistical significance—2.5 times greater risk on basic analysis, with significance lost after adjusting for sample size. Still, we believe that the pancreas as the primary site is itself an important prognostic factor. Another important factor is final disease stage. It is well known that higher stages are associated with higher probability of recurrence. Our analysis revealed an (unadjusted) OR of 1.3 for relapse associated with disease stage. However, we feel that the near equal distribution of cases among all three stages compromised the sample size, and the same was represented in the confidence intervals.

We also compared risk of recurrence in each stage after adjusting for the presence or absence of complications. In the final analysis, we found no differences at any given stage between patients who had complications and those who did not. Therefore, stage had an influence on recurrence regardless of complication status. We believe that these two factors—primary site and stage—require larger, more focused study before conclusions can be drawn regarding influence on risk of recurrence.

Due to limited patient numbers, this study was likely insufficiently powered to demonstrate conclusively the ability of anastomotic leakage, delayed recovery, or excessive weight loss to predict relapse-free survival in patients who have undergone definitive surgery for gastrointestinal malignancies. However, it certainly raises a question for researches as to how greater-than-expected postoperative weight loss correlates to poor overall outcome. For clinicians, it also underscores the importance of avoiding anastomotic leakage when planning postoperative management of this patient population.