The clinical signs of diverticulitis and colon cancer are known to overlap, both in history and on imaging, and the two entities may be difficult to distinguish from each other in certain patients. Approximately 2% to 3% of patients with suspected acute diverticulitis ultimately have a diagnosis of colon cancer, with higher rates in patients with complicated diverticular disease.
Avoiding this misdiagnosis is critical to successful cancer control. Treatment of colon cancer requires a complete staging workup and, unless metastatic, prompt surgical treatment according to the principles of oncologic resection. On the contrary, due to its benign condition, diverticulitis can be treated surgically or conservatively, depending on the presentation, and surgery does not need to add lymph node removal.
Historical recommendations advocating resection after 2 episodes have been replaced by clinical practice guidelines recommending observation of uncomplicated acute diverticulitis and selected complicated cases of diverticulitis with abscess . With this paradigm shift, if the diagnosis of diverticulitis is not ensured, there is the possibility of leaving untreated colon cancer in situ.
In this study, the authors hypothesize that there are preoperative patient characteristics associated with an unexpected diagnosis of malignancy in the setting of suspected acute diverticulitis that may help predict cancer in this setting. The objective was to identify a set of preoperative variables associated with unexpected malignancy in a large national cohort.
| Methods |
> Study design
This is a retrospective cohort study of data from the American National Surgical Quality Improvement Program (ACS-NSQIP). ACS-NSQIP is a database focused on variables related to 30-day postoperative outcomes.
Beginning in 2012, a colectomy data set comprising 23 specificity variables for colon surgery was collected and made available. These data include information on colon cancer staging. All data available at the time of this study (2012-2018) were used.
> Patient selection and definitions
Patients were included if they had a primary surgical indication of “acute diverticulitis.” Patients with unexpected cancer were identified within this group if they had pathologic staging information recorded as T1 or N1. Both emergency and elective cases were included.
> Statistical analysis
Descriptive statistical analyzes (means, frequencies, etc.) were performed to characterize the study population. Variables of interest included age, sex, race, preoperative body mass index, preoperative weight loss (defined as loss of > 10% of body weight in 6 months), preoperative albumin, preoperative hematocrit, perioperative blood transfusion (> 1 unit within 72 hours), preoperative white blood cell count, preoperative length of stay (LOS) > 1 day, emergency operation, and urgent surgical indication.
For cancer patients, variables related to staging (T stage, N stage, TNM classification) were analyzed. Markers of inappropriate cancer surgery included poor lymph node yield (<12) or positive margin. Multivariate logistic regression models were constructed to estimate associations between unexpected cancer and variables that were significant while adjusting for potential confounders. The threshold for statistical significance was set at P < 0.05.
| Results |
The authors identified 17,368 patients with a preoperative diagnosis of diverticulitis. Cancer was diagnosed postoperatively in 164 patients (0.94%) .
Compared with patients with a postoperative diagnosis of benign disease , patients with unexpected cancer were older ( 65.7 ± 15.4 vs. 60.6 ± 13.7 years), with higher rates of anemia (25. 6% vs. 17.6%) and sepsis (43.9% vs. 29.9%).
Patients with unexpected cancer also had higher incidences of weight loss (11.6% vs. 3.8%) and lower preoperative albumin levels (3.11 ± 0.8 vs. 3.48 ± 0.8 g /dl).
Regarding cancer staging for those patients with malignancy, tumors were locally advanced in 84% of cases (39% T3, 45% T4) and approximately half of patients did not have lymph node metastases (55 .5% no). The most common TNM classification was II (43.9%) followed by III (38.4%). Eight patients had findings of distant metastasis (4.9%). 17% of patients with unexpected malignancy had inadequate oncologic resection.
Of the significant differences between unexpected cancer and diverticulitis identified in the bivariate analyses, the odds ratios (OR) for sepsis (OR 2.14, 95% confidence interval [CI] [1.3, 3.6 ]; P < 0.01), weight loss (OR 2.31, 95% CI [1.1, 4.4]; P ¼ 0.01), and preoperative albumin level (OR 0.64 , 95% CI [0.45, 0.92]; P < 0.01) remained statistically significant after adjustment for covariates. Sepsis was correlated with more advanced T stage (P = 0.01).
| Discussion |
This study examined patients with unexpected diagnoses of malignancy after surgery for suspected acute diverticulitis using a large national database. There are several key findings.
First, the diagnosis of an unexpected cancer in this particular population was rare. Patients diagnosed with cancer were older and had a higher incidence of anemia, sepsis, and malnutrition than other patients with suspected acute diverticulitis.
After adjusting for confounders, only preoperative sepsis and malnutrition remained statistically associated with cancer diagnosis.
Patients with sepsis or recent weight loss had a more than twofold increase in the odds of malignancy. For each increased unit (1 g/dL) of albumin, patients had a 36% decrease in the odds of malignancy. Together, these findings may help inform the surgeon and patient about the increased likelihood of cancer preoperatively.
This information may be instrumental in assisting with key decisions related to surgical treatment, including nonsurgical versus operative strategies, timing of surgery, and choice of operation. The additional finding that 17% of patients with unexpected malignancy had inadequate oncologic resection highlights the need to improve preoperative diagnosis in this setting.
The relationships between cancer and (1) sepsis at presentation, (2) preoperative hypoalbuminemia , and (3) preoperative weight loss remained strong after adjusting for clinical presentation with all of the aforementioned variables. Therefore, these 3 predictors may be associated with the presence of malignancy in cases of suspected acute diverticulitis , regardless of the clinical presentation.
Studies have shown that patients recovering from complicated diverticulitis have a significantly increased risk of advanced adenoma or cancer, while patients with uncomplicated diverticulitis do not have an increased risk compared to the general population.
In updated recommendations from the American Society of Colon and Rectal Surgeons, colonoscopic evaluation after acute diverticulitis is reserved for patients with worrisome symptoms (eg, narrow stools, bleeding), aberrant recovery, worrisome imaging findings such as mesenteric/retroperitoneal abscess and/or obstruction, or lymphadenopathy.
Data from the present study suggest that malnutrition and sepsis are additional indicators of increased cancer risk, and that patients with these characteristics during their initial presentation may also warrant a colonoscopy, assuming they do not require initial emergency surgery.
Malnutrition and weight loss are known to be associated with cancer and therefore the related observation in this study is in accordance with previous findings. Multiple studies have also shown that malnutrition is specifically associated with colorectal cancer.
It is important to recognize that preoperative serum albumin may be an inaccurate measure of overall nutritional status, as serum albumin may also decrease due to systemic inflammation. The weight loss data from the current study are more correlated with malnutrition and suggest that patients with suspected diverticulitis who are malnourished have a higher likelihood of findings of malignancy.
The association between sepsis and malignancy was not expected because sepsis is a feature traditionally associated with diverticulitis. However, there are previous data demonstrating correlations between colorectal malignancy and sepsis.
Although sepsis is not a direct cause of cancer, it is possible that the 2 conditions share similar risk factors ; specifically, decreased immune function, underlying systemic inflammation, or prolonged antibiotic therapy could be associated with cancer through of multiple mechanisms.
The fundamental principles of sigmoid colectomy for diverticulitis are resection distal to the rectosigmoid junction and proximal to the healthy colon that is suitable for anastomosis. This is quite different from oncologic colectomies, which involve en bloc removal of the tumor with negative margins of prescribed length, as well as adequate removal of lymph nodes (12-15 lymph nodes).
The current study showed a positive margin rate of 1.2% and inadequate lymph node harvesting of 15.9%. Positive margin is a poor prognostic feature with reduced overall survival compared to those with negative margins. However, the positive margin rate of 1.2% in this cohort is not high compared to the general literature.
Lymph node status is an important prognostic factor in colorectal cancer and significantly affects survival.
In the current study, 16% of patients diagnosed with incidental carcinoma have inadequate lymph node performance. In part, this unexpected result could be due to a high rate of missing data regarding lymph node status in this study (21%).
According to National Comprehensive Cancer Network guidelines, inadequate oncologic resection requires additional interventions, such as chemotherapy and consideration of boost radiation. The findings of the current study provide insight into which patients with suspected diverticulitis have a higher likelihood of malignancy and, therefore, who would benefit from oncologic resection.
This study has several strengths worth mentioning. Use of the ACSNSQIP targeted colectomy data allowed an in-depth account and review of preoperative characteristics related to potential cancer risk. The cohort size was adequate for meaningful regression analysis with adjustment for multiple potential confounders. The study findings also help answer a question that is highly relevant to current clinical practice.
Possible conclusions from the study findings are limited by the retrospective nature of the analysis and the source data. ACSNSQIP does not include certain variables that may be related to cancer risk, such as data related to prior colonoscopy findings, family or personal history of colorectal cancer, coincident diagnosis of inflammatory bowel disease, or prior intra-abdominal radiation. This introduces the possibility of unexplained confusion.
It also does not include more specific pathological data, such as pathological diagnosis, or high-risk histological characteristics. These details would have helped provide information about the biology of the underlying cancers, but would not have altered the main findings of the study. There was also no way to definitively determine from the data set whether the operating surgeon had a clinical suspicion of cancer preoperatively or whether the tumors found on postoperative pathology were truly incidental.
An additional limitation of the data is the inability to reconcile why a number of cancer patients listed as perforated had a tumor staging
| In conclusion , an unexpected postoperative diagnosis of cancer occurs in a small percentage of patients with suspected diverticulitis. Surgeons should have a high index of suspicion for cancer in patients with sepsis or malnutrition. Although most patients will ultimately undergo adequate oncologic resection regardless, surgeons must be especially attentive to resection in accordance with oncologic principles when managing patients with these features. If such cases are suitable for elective surgery, there should be a low threshold for performing preoperative colonoscopy because there is a higher probability of malignancy. |
