Colorectal and Extracolonic Cancers Detected at Screening CT Colonography in 10 286 Asymptomatic Adults

Introduction

The primary target for colorectal carcinoma (CRC) screening and prevention is advanced neoplasia, which includes both CRC itself, as well as benign but histologically advanced adenomas that are at increased risk for progression to malignancy (1,2).Prevention of CRC through detection of advanced adenomas with complete structural examination, either with optical colonoscopy or computed tomographic (CT) colonography, is a preferred screening strategy, as recognized by the recent guideline recommendation from the American Cancer Society, in conjunction with the major gastroenterology and radiology societies (3). Early detection of presymptomatic invasive CRC, although less frequently identified than advanced adenomas in a screening population, is clearly an important goal as well.

In addition to the intended colorectal evaluation, CT colonography also provides for limited assessment of extracolonic structures, including the abdomen, pelvis, and lung bases. Additional diagnostic imaging work-up for unsuspected extracolonic findings is performed in about 6% of cases, nearly half of which ultimately prove to be clinically relevant (4). The purpose of this study was to retrospectively determine the detection rates, clinical stages, and short-term patient survival for all unsuspected cancers identified at screening CT colonography, including both CRC and extracolonic malignancies.

Materials and Methods

P.J.P. and D.H.K. have consulted for Viatronix and Medicsight and are cofounders of VirtuoCTC. P.J.P. has also consulted for Covidien & Philips.

This Health Insurance Portability and Accountability Act-compliant study was performed at two medical centers by using institutional review board-approved protocols. Outpatient screening CT colonography was performed in 10286 asymptomatic adults (mean age, 59.8 years) that included 5388 men (mean age, 60.0 years) and 4898 women (mean age, 59.7 years) from April 2004 through March 2008. Standard bowel preparation prior to CT colonography consisted of a saline laxative (magnesium citrate or sodium phosphate), dilute barium, and diatrizoate, as previously described, with only slight variations in the specific protocols (2,5). Colonic distention for CT colonography was achieved with automated carbon dioxide distention in the majority of cases and with manual room air insufflation in the initial minority of cases. The low-dose, nonenhanced multidetector CT protocol has also been described previously (2,5).

Prospective evaluation for both colorectal and extracolonic disease was performed by one of seven experienced radiologists (including P.J.P., D.H.K., D.S.B.) immediately after the examination was completed to facilitate same-day polypectomy, if needed. The radiologists were all experienced in CT colonography interpretation, having read more than 500 cases. The CT Colonography Reporting and Data System, or C-RADS, was used prospectively for both colorectal and extracolonic findings in all cases (6). A dedicated CT colonography software system (V3D Colon; Viatronix, Stony Brook, NY) was used for the colorectal evaluation, with reporting of all potential lesions measuring 6 mm or greater in size. With the exception of a subset of patients with 6–9-mm polyps placed in a short-term CT colonography surveillance, cases positive for polyps 6 mm or greater were further evaluated at optical colonoscopy. For the purposes of this study, we focused on patients with a pathologically proved invasive CRC, defined as spread of malignant cells beyond the muscularis mucosae. By definition, advanced adenomas with or without high-grade dysplasia, although a critical target for CRC prevention, were not included in this cancer analysis (2). Furthermore, patients with a known or suspected of having colorectal malignancy based on the presence of signs or symptoms were excluded from consideration.

Prospective extracolonic evaluation was performed by the interpreting radiologist in conjunction with the indicated colorectal evaluation per routine. For cases in which further work-up was recommended for an unexpected but potentially important extracolonic finding, retrospective electronic medical record review was performed to document all clinically unsuspected and pathologically confirmed extracolonic cancers. If further work-up was performed outside of the participating centers and was not available by means of a record review, the referring provider was contacted to supply the necessary information. Possible cancers without histopathologic diagnosis were excluded, including cases where the patients were undergoing imaging surveillance (eg, small renal masses in elderly patients). For any unresolved cases, periodic chart reviews were performed, including consultation with the referring physician as indicated. As with cases of symptomatic CRC, any patient with a known or suspected of having extracolonic cancer based on symptoms was excluded from further consideration. In addition, patients with a diagnosis of cancer at any point after CT colonography evaluation were excluded if there was no associated finding that was prospectively identified at CT colonography.

For all patients with a pathologically proved unsuspected cancer detected at CT colonography, additional diagnostic work-up, method of tissue diagnosis, clinical tumor stage, mode of therapy, and subsequent clinical outcome (assessed from clinical oncology follow-up notes, medical records review, and direct provider feedback) were assessed and tabulated. TNM staging was used, except for cancers where an alternate staging classification is generally used (eg, lymphoma and carcinoid tumor).

For statistical analysis, primary comparisons were made between patients with unsuspected cancers detected at CT colonography and the remainder of the screening population. Continuous variables were compared between two groups by using a two-sample z test, while categorical variables were compared by using Fisher exact test. The cancer detection rate between the two centers was compared by using Fisher exact test. A two-sided P value of less than .05 was considered to indicate statistical significance (SAS, Cary, NC).

Results

After complete clinical work-up of newly detected findings at screening CT colonography, unsuspected malignancy was pathologically confirmed in 58 (0.56%) of the 10286 subjects, including CRC in 22 (0.21%) patients (Table E1 [online]) and extracolonic cancer in 36 (0.35%) patients (Table E2 [online]). The mean age of these 58 patients with unsuspected cancer was 60.8 years, which was only 1 year older on average compared with the mean age of those without cancer (59.8 years, P = .32). Women comprised a slight majority of unsuspected cancer cases (56.9%) despite being a slight minority among the entire screening cohort (47.6%) (P = .13). Cancer detection rates were similar at both medical centers (28 of 5576 versus 30 of 4710; P = .41).

The clinical extent of disease at diagnosis was stage I or localized in 31 (53.4%) cases. Stage II, III, and IV disease was present in six, 12, and seven cases, respectively, in addition to regional lymph node involvement in two patients with carcinoid tumor. A total of 46 (79.3%) patients underwent surgical (n = 43) or endoscopic (n = 3) resection of tumor; two patients with proved renal cell carcinoma underwent percutaneous ablation (one cryoablation, one microwave ablation). A total of 19 (32.8%) patients received some form of chemotherapy or hormonal therapy, and seven (12.1%) patients received radiation therapy.

The mean time interval for clinical follow-up was 30.2 months ± 11.8 (standard deviation), with a range of 12–56 months. The follow-up interval was at least 1 year for all patients. At the time of the most recent follow-up, 54 (93.1%) of 58 patients were alive, with the majority in good health and showing no evidence of disease. To our knowledge, in no patient who presented with stage I, stage II, or localized disease at diagnosis has the disease progressed to a higher stage. Deaths in three patients (5.2%) were related to the detected cancer (metastatic CRC, metastatic bronchogenic carcinoma, and malignant mesothelioma), all of which were advanced at diagnosis; one other patient died of an unrelated cause.

Invasive Colorectal Cancer

Specific findings for the 22 asymptomatic patients with unsuspected invasive CRC detected at CT colonography are listed in Table E1 (online). The mean age was 60.0 years, which was not significantly different from that in the general screening cohort (P = .80) and included 14 women and eight men. With the exception of a malignant cecal carcinoid tumor, all of the remaining tumors were typical colorectal adenocarcinomas (Fig 1). All but one patient underwent optical colonoscopy for tissue diagnosis prior to definitive treatment, most of which was performed the same day as CT colonography. One 50-year-old woman with a mass in the ascending colon also had a sigmoid stricture from unsuspected endometriosis, which precluded endoscopic evaluation of the proximal tumor. Stage I CRC was confirmed at surgery in this patient.

Figure 1a:

Download as PowerPointOpen in Image Viewer

Figure 1b:

Download as PowerPointOpen in Image Viewer

Figure 1c:

Download as PowerPointOpen in Image Viewer

The tumors were typically large (mean size, 3.3 cm ±1.6; range, 1.1–8.0 cm), appearing as either frankly invasive masses or malignant polyps. Segmental tumor location was sigmoid colon in eight cases, cecum in five cases, transverse colon in two cases, rectum in three cases, ascending colon in three cases, and descending colon in one case. One patient had two synchronous cancers in the transverse colon. Most patients had stage I disease (n = 12, 54.5%), three patients stage II disease, four patients had stage III disease, and two patients had stage IV disease at diagnosis. The case of cecal carcinoid had regional nodal spread. All patients with CRC underwent resection of the primary tumor, with six patients also receiving chemotherapy and two receiving radiation therapy. One patient refused additional therapy after surgery and was lost to follow-up after 17 months. One patient who presented with asymptomatic stage IV CRC died 12 months after diagnosis; the remaining patients are alive as of the last clinical follow-up.

Extracolonic Cancer

Specific findings and tumor types for the 36 patients with unsuspected extracolonic cancer detected at CT colonography are listed in Table E2 (online). The mean age was 61.3 years, which was not significantly different from that in the general screening cohort (P = .36). There were 19 women and 17 men, also not significantly different from the entire group (P = .49). The most common extracolonic primary tumors were renal cell carcinoma (n = 11), lung cancer (n = 8), and non-Hodgkin lymphoma (n = 6) (Fig 2). As of the last clinical follow-up, only one patient had died of an unrelated cause.

Figure 2:

Download as PowerPointOpen in Image Viewer

Of the eight patients in whom lung cancer was diagnosed, seven were non–small cell type (adenocarcinomas) and one was malignant mesothelioma. All tumors were detected at the visualized lung bases, since the mid and upper lung zones are not imaged at CT colonography. Three patients had stage I disease and five had stage III disease. Seven patients underwent surgery, five received chemotherapy, and five received radiation therapy. All patients were alive at the most recent follow-up, except for one patient with bronchogenic carcinoma and the one patient with malignant mesothelioma.

Among the 11 patients with a diagnosis of renal cell carcinoma, stage I disease was diagnosed in 10. Of these nine patients with localized disease, eight underwent surgical resection and two underwent percutaneous ablation. The remaining patient had stage IV renal cell carcinoma, characterized by several small pulmonary nodules. This patient did well for 27 months but then died of an unrelated cerebrovascular cause. The other nine patients were alive and well without evidence of tumor recurrence at most recent follow-up.

Three patients with large adrenal tumors underwent surgical resection. At pathologic evaluation, two cases proved to be pheochromocytomas and the other proved to be stage I adrenocortical carcinoma (Fig 3). All three tumors were localized with negative surgical margins. In retrospect, one patient with adrenal pheochromocytoma had a history of heart palpitations and headaches that were presumed to be migrainous in nature. All three patients were alive and well at most recent follow-up.

Figure 3:

Download as PowerPointOpen in Image Viewer

In three patients, unsuspected metastatic disease from previously treated localized cancer that was believed to be cured was diagnosed. One patient with a history of stage I endometrial carcinoma had unsuspected peritoneal implants detected at CT colonography. In a second patient with a history of stage I breast cancer, unsuspected liver metastases were detected at CT colonography. A third patient had mesenteric lymphadenopathy from a previously excised cutaneous malignancy (squamous cell carcinoma). All three patients were alive at most recent follow-up.

Discussion

This study was performed to assess the detection rate and clinical outcome of unsuspected malignancies detected with CT colonography in an asymptomatic screening population. Our findings show that unsuspected cancer is found on average more than once every 200 adults screened, and most are extracolonic malignancies. At first glance, it may seem paradoxical that extracolonic cancers outnumber invasive CRC given the indication and focus of the examination. However, CRC was still the predominant cancer identified, followed by renal cell carcinoma, bronchogenic carcinoma, and non-Hodgkin lymphoma. Furthermore, one must keep in mind that CRC prevention through detection and removal of large advanced adenomas is the primary goal of screening colonoscopy, whether by virtual or conventional means (3). Large polyps (≥ 10 mm) are detected in 4%–5% of asymptomatic adults undergoing CT colonography in our screening programs, corresponding to nearly 500 patients in this cohort. The identification of these large polyps likely represents the single greatest benefit of this screening test (2,7).

Our results suggest that early detection of asymptomatic extracolonic cancer represents an additional benefit of screening CT colonography that is not available with alternative screening options such as optical colonoscopy. However, it is important to consider the potential impact of biases that may cause overestimation of detection rates and survival, including lead time bias, length bias, and overdiagnosis bias. Lead time bias refers to overestimation of survival duration measured from the time of diagnosis for screen-detected cases relative to patients presenting with signs and symptoms. This is a potentially minor contributor since most patients in this series presented with early-stage cancer and are presumably cured of their disease. Length bias results from the relative excess of slowly progressing cases, which are disproportionally identified at screening due to a longer window for detection. Length bias is likely a factor in some slow-growing renal cell carcinomas but perhaps less of a factor in other cases, such as bronchogenic carcinoma. Overdiagnosis bias refers to the inclusion of subclinical disease or pseudodisease that would not become overt before the patient dies of other causes. For example, some cases of renal cell carcinoma may effectively represent cancer without disease (8).

Potential biases aside, a disproportionate number of the asymptomatic screening-detected cancers in our series were localized or at an early stage, which appears to have positively affected the expected short-term survival. There is abundant evidence that cancer survival is largely related to the stage of disease and that symptomatic cancers tend to present at a more advanced stage. For example, according to the Surveillance, Epidemiology, and End Results (SEER) cancer statistics review (9), the 5-year survival for localized CRC is 90% but fewer than 40% of patients receive a diagnosis at this stage. In comparison, the majority of CRC in our series was identified as localized or stage I disease. For regional and distant CRC, 5-year survival decreases to 68% and 10%, respectively. Likewise, the 5-year survival for localized renal cancer is 90% but only 55% of patients receive a diagnosis at this stage (compared with >90% of patients in our series). Survival decreases to 10% for renal cancer with distant spread. For non–small cell lung cancer, the 5-year survival is 51% for localized disease, but over 80% of patients have more advanced disease at diagnosis according to the SEER, and the 5-year survival for lung cancer with distant spread is only 3%. In our series, 42.9% of patients with non–small cell lung cancer received a diagnosis of asymptomatic localized disease as a result of undergoing screening CT colonography.

Beyond the cancers that were identified at screening CT colonography in this cohort, there is a conspicuous absence of certain extracolonic malignancies. Most notably, we did not prospectively identify any cases of presymptomatic pancreatic or ovarian cancer, which probably relates to the relatively narrow window of time available for detection of asymptomatic disease for these tumors, which often progress rapidly and typically manifest as advanced. In addition, the low-dose, nonenhanced technique utilized for screening CT colonography limits the detection of pancreatic tumors.

Inclusion of CT colonography as a recommended test in the recently revised screening guidelines from the American Cancer Society may ultimately lead to broader third-party reimbursement and more widespread implementation (3). At the time of this writing, 23 states and the District of Columbia mandate coverage for screening tests endorsed by the American Cancer Society, including CT colonography. Beyond issues related to colorectal evaluation, the potential impact from extracolonic findings also becomes relevant from both a clinical efficacy and cost-effectiveness standpoint. In practice, extracolonic evaluation at CT colonography represents a double-edged sword (10), with potential benefits related to the reassurance that important disease is absent in the great majority of patients and to the preclinical detection of an unsuspected but clinically important disease process in a small minority. Potential harms include the anxiety, inconvenience, potential complications, and added costs related to additional diagnostic work-up for findings that ultimately prove to be unimportant. On the basis of a number of clinical studies, extracolonic findings of at least moderate potential importance are found at CT colonography in 7.4%–11.4% of cases (4,11–13). In our experience, 6.1% of patients screened with CT colonography undergo additional evaluation for unsuspected extracolonic findings, and a relevant new diagnosis will be made in 2.5% (4). Most of the costs related to extracolonic work-up are for truly relevant diseases that require invasive diagnosis and/or treatment.

When considering the cost-effectiveness of screening with CT colonography versus optical colonoscopy, it is important to consider both costs and benefits related to extracolonic evaluation at CT colonography, as well as the potential complications related to each screening procedure. When detection of extracolonic cancers and aortoiliac aneurysms is included along with CRC screening, CT colonography has been shown to be more clinically effective and more cost-effective than optical colonoscopy (14).

There were a number of limitations to our study. The potential biases inherent to screening have already been discussed. The duration of longitudinal follow-up averaged only 2.5 years. However, the early stage of most cancers at screening detection portends a more favorable prognosis and presumably complete cure in many cases. Some presumptive cancers were excluded from consideration because histopathologic confirmation was lacking, such as patients with likely renal cell carcinomas undergoing watchful waiting without biopsy. Last, some tumors included in this series, such as adrenal pheochromocytomas, were localized and it could be argued that they represent benign disease and not true cancers at this stage. However, we also detected one case of stage I adrenocortical carcinoma, which is found at this early stage in only 3% of cases (15). We did not include advanced colorectal adenomas with high-grade dysplasia, which have been referred to as “carcinoma in situ” in the past, with CRC in this series. In this article we did not specifically address the issue of cancer prevention through detection and removal of advanced adenomas, but this has been reported elsewhere (2).

In conclusion, clinically unsuspected cancer was detected at routine screening CT colonography of generally healthy adults with a frequency of greater than one case per 200 individuals screened. Although invasive CRC was the most common malignancy overall, most detected cancers were extracolonic in location. Presymptomatic detection of unsuspected colorectal and extracolonic malignancy resulted in a high rate of localized or early-stage tumors, with a very favorable clinical outcome to date.

Author Contributions

Author contributions: Guarantors of integrity of entire study, P.J.P., B.D.C.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, P.J.P., R.J.M., R.A.R., B.D.C.; clinical studies, P.J.P., K.S.W., M.E.H., D.S.B., P.A.C., B.D.C.; experimental studies, K.S.W., B.D.C.; and manuscript editing, P.J.P., D.H.K., R.J.M., M.E.H., R.A.R., B.D.C.