What Is the Clinical Importance of Incidental Findings on Staging CT Scans in Patients With Sarcoma?

Abstract

Background

Baseline staging CT scans are performed on nearly every patient after the diagnosis of a sarcoma to evaluate for the presence of metastatic disease. These scans often identify abnormalities that may or may not be related to the known malignancy. Despite the high frequency of incidental findings, there is little guidance for clinicians faced with assessing these radiographic abnormalities. The interpretation of incidental findings is important because it may influence decisions regarding surveillance frequency, prognostic estimation, and surgical and medical intervention.

Questions/purposes

The purpose of this study was to determine (1) the frequency of abnormal findings and indeterminate nodules on staging CT scans; (2) the natural history of indeterminate nodules identified at the time of sarcoma diagnosis; and (3) the factors associated with indeterminate nodules representing true metastatic disease.

Methods

Between September 2010 and February 2016 we treated 233 patients with bone and soft tissue sarcomas. Of those, 227 (97%) had a staging CT scan of the chest or chest/abdomen/pelvis performed within 2 months of diagnosis. To be eligible for this retrospective study, a patient had to have a minimum of 6 months of radiographic followup after that initial CT scan. A total of 36 (16%) were lost to followup or did not have radiographic surveillance at least 6 months later, and 48 (21%) were excluded for other prespecified reasons, leaving 149 patients for evaluation. We recorded all abnormal findings listed in the official radiology CT report of the lung, bone, liver, and lymph nodes. We assessed progression of indeterminate nodules by reviewing radiology reports, which listed both size and number of findings, and clinical notes outlining the current assessment of disease status and treatment plan. If indeterminate nodules grew in size or number consistent with metastatic disease or were confirmed histologically, they were considered to represent true metastasis. Bivariate methods were used to investigate an association between various clinical factors, which were obtained from chart review, and progression of indeterminate nodules to clear metastatic disease.

Results

One hundred thirty-five of 149 patients (91%) had at least one abnormal finding on a staging CT scan. Forty-nine patients (33%) presented with indeterminate lung nodules, 15 (10%) with indeterminate liver lesions, four (3%) with indeterminate bone lesions, and 57 (38%) with enlarged lymph nodes. Fifteen of the 49 patients with indeterminate lung nodules (31%), one of 15 liver nodules, zero of four bone lesions, four of 13 lymph nodes 1 to 2 cm in size, and two of 44 subcentimeter lymph nodes (4.5%) were clearly metastatic on followup. A primary tumor size ≥ 14 cm in greatest dimension was more suggestive of indeterminate nodules representing true metastatic disease compared with smaller primary tumors in both lung (eight of 10 compared with seven of 36 [19%]; odds ratio, 16.6; 95% confidence interval, 2.9-95.9; p < 0.001) and lymph nodes (six of 18 compared with zero of 36 [0%], p < 0.001).

Conclusions

It is extremely common for abnormal findings and incidental nodules to be present at the time of a staging CT scan in patients with sarcoma. Although patients with indeterminate nodules should have continued surveillance, it appears from this study that the majority of these findings do not represent true metastatic disease. Given a minimum followup of 6 months, it is possible the actual proportion of indeterminate lesions representing true metastatic disease may increase over time.

Level of Evidence

Level III, prognostic study.

Introduction

According to the National Cancer Institute, approximately 12,000 patients with soft tissue sarcoma and 3000 patients with bone sarcoma will be diagnosed in the United States annually [18]. When the diagnosis of a sarcoma is made, one of the first imaging studies obtained is a staging CT scan of the chest alone or of the chest, abdomen, and pelvis [8, 14]. The purpose of these scans is to look for distant sites of disease, most notably pulmonary metastases, because the lung is the most common site of metastasis in patients with both soft tissue and bone sarcoma [1, 6-8, 17, 21]. Many times the initial imaging studies will report additional findings, which may or may not be related to the known malignancy [14, 20, 21]. The clinician then must determine if these additional findings are nonneoplastic or benign and unrelated to the underlying sarcoma, are related to the underlying malignancy, or represent an important additional medical condition that should have further intervention.

Two recent studies have been conducted on the incidence of incidental findings at staging CT scan in prostate and colorectal cancer [10, 15]. However, there are few reports on the frequency of incidental findings at the time of staging CT scan in patients with sarcoma [14, 20]. Indeterminate nodules or findings of uncertain importance in the lung, liver, bone, and lymph nodes are commonly seen on staging examinations for sarcoma and may indicate the presence of metastatic disease, unrelated benign lesions, or nonneoplastic coincident conditions [1-4, 6-9, 11, 17, 19-22, 24]. Because the detection of indeterminate nodules is becoming much more common with the constantly increasing performance of CT scans [12, 17], and nearly all patients with sarcoma will be evaluated for systemic disease with a CT scan, determining the underlying biology and importance of incidental findings is a clinical challenge. It is important that the treating team of physicians have all available information needed to help make management decisions regarding these indeterminate lesions. In particular, the treating team often has to make a decision about whether to biopsy a suspicious finding on CT or periodically monitor by repeat imaging. Currently, there are limited criteria for making that decision.

The purpose of this report was to determine (1) the frequency of abnormal findings and indeterminate nodules on staging CT scans; (2) the natural history of indeterminate nodules identified at the time of sarcoma diagnosis; and (3) the factors associated with indeterminate nodules representing true metastatic disease.

Patients and Methods

This study was an institutional review board-approved single-institution retrospective chart study of 233 potentially eligible patients diagnosed with a primary sarcoma of the extremities from September 2010 to February 2016. We included all histologic subtypes, grades, and tissues of origin (bone and soft tissue). We excluded atypical lipomatous tumors and any patients with a history of sarcoma or recurrence of a previously treated sarcoma. All patients had a chest (39 patients) or chest, abdomen, and pelvis CT scan (110 patients) performed within 2 months of diagnosis. A radiologist at our institution formally read each of these CT scans with a corresponding report in the medical record. These official reports were used to record all reported abnormalities on baseline imaging. Patient records, including clinic notes and pathology reports, were reviewed to determine underlying patient and tumor characteristics. Our protocol is to monitor patients with imaging surveillance of the chest every 3 to 6 months for the first 2 years followed by every 4 to 12 months for years 3 to 5, depending on tumor grade, histology, and specific individual clinical factors. Patients with indeterminate nodules were required to have a minimum of 6 months of radiographic (CT) followup to allow for determination of progression or stability. Patients who were treated with perioperative chemotherapy were excluded if they were not monitored for at least 6 months after completing systemic treatment.

Of the 233 originally identified patients, 227 (97%) had a staging CT scan of the chest or chest/abdomen/pelvis performed within 2 months of diagnosis. Overall, 84 were excluded. No patients had a prior thoracotomy in the area of a suspicious lesion. Reasons for exclusion included atypical lipomatous tumors (17), recurrent disease (17), staging CT scans done externally and not read by our radiologists (eight), staging CT scans not performed within 2 months of diagnosis (six), and < 6 months of documented followup CT scans (36). These 36 patients lost to followup had their initial staging CT scan read at our institution, but died, were followed at a different institution, or had imaging followup that consisted of plain radiographs rather than CT scans.

The final cohort consisted of 149 patients with a median age of 51 years (range, 5-95 years). Male patients (79) were more frequent than females (70), and soft tissue sarcomas (92) were more frequent than bone sarcomas (57). There were 19 different variants of sarcoma included in this study; osteosarcoma was the most common histologic subtype (25) followed by myxofibrosarcoma (24), undifferentiated pleomorphic sarcoma (24), chondrosarcoma (22), Ewing’s sarcoma (10), leiomyosarcoma (nine), malignant peripheral nerve sheath tumor (seven), liposarcoma (seven), synovial sarcoma (six), epithelioid sarcoma (four), rhabdomyosarcoma (two), spindle cell sarcoma (two), and one each of alveolar soft parts sarcoma, clear cell sarcoma, dermatofibrosarcoma protuberans, malignant glomus tumor, myxoid sarcoma, sclerosing epithelioid fibrosarcoma, and myoepithelioid sarcoma. Eighteen patients had Grade 1 sarcoma (12%), 17 had Grade 2 (11%), and 114 (77%) had Grade 3. Chest, abdomen, and pelvis staging scans (110) were more common than staging scans of the chest alone (39). Patients who did not develop metastatic disease had a median of seven CT scans (range, 2-17) over a mean followup of 24 months (range, 6-71 months).

The primary outcome for staging CT imaging was the presentation of lung, bone, liver, or lymph node metastases related to the known sarcoma. All nodules, masses, or other abnormalities noted on the radiology report of the staging CT examination were classified as metastatic sarcoma, indeterminate (possibly metastatic sarcoma), or not metastatic sarcoma. Nodules were considered metastatic at presentation if they were confirmed histologically or if the appearance was highly suspicious for metastatic disease. We recorded all abnormalities listed for each patient, which could include more than one site. We confirmed that the lesions considered metastatic at presentation that were not biopsied demonstrated behavior consistent with metastatic disease, specifically, continued growth and development of additional lesions on subsequent scans. Lung nodules were considered metastatic at presentation if multiple nodules were present with at least one nodule ≥ 1.0 cm. Bone lesions > 2 cm that involved multiple sites of the skeleton or caused bony destruction with a soft tissue mass were considered metastatic at presentation. Lymph nodes > 2 cm in conjunction with clear diffuse pulmonary or osseous disease were considered metastatic at presentation. Nodules for each location were considered indeterminate if the appearance was not clearly suggestive of a specific process such as a granuloma, cyst, or hemangioma and could be representative of early metastatic sarcoma. The number of nodules as well as the size of the largest nodule was recorded. Clinic notes and imaging review by the authors (ZM, BJM) were used if the radiology report was unclear in its description. Indeterminate nodules were considered benign if they remained stable or resolved on subsequent surveillance scans. If the indeterminate nodules grew in size or number, consistent with metastatic disease, or if they were confirmed histologically, they were noted to represent true metastatic disease.

Finally, we compared patients with indeterminate nodules that progressed to metastatic disease of the lungs, liver, bone, or lymph nodes with patients with indeterminate nodules that did not progress to metastatic disease in these locations. Bivariate methods (chi-square and Fisher’s exact testing) were used to investigate the association of various clinical characteristics with metastatic progression. Factors we considered included age, sex, race, body mass index, age-adjusted Charlson Comorbidity Score, smoking history, bone versus soft tissue sarcoma, histology, location, depth, grade (Grade 1 versus Grade 2 and 3), tumor size, indeterminate nodule number, indeterminate nodule size, and indeterminate nodules present in multiple locations. We conducted many analyses of sizes (for example, > 5 cm versus < 5 cm, > 10 cm versus < 10 cm) as well as treating the number as a continuous variable. A tumor size of 14 cm showed significant differences compared with other sizes. Our biostatistician (YG) felt this was a reasonable way to determine a critical size. Tumor size could not be assessed in three patients, because they had an initial excision performed before referral to our institution without clear documentation of tumor size before excision.

Results

Frequency of Abnormal Findings

Overall, of the 149 patients included in this study, 135 (91%) had at least one abnormal finding at the time of staging CT scan (Table 1). Fourteen patients (9%) demonstrated metastatic disease, 80 patients (54%) had an indeterminate nodule, and 55 patients (37%) had CT scans without lesions. The highest incidence of indeterminate nodules was in the lung (49 patients [33%]) followed by the liver (15 [10%]) and bone (four [3%]). Thirteen patients (9%) had at least one indeterminate lymph node ≥ 1.0 cm. Thirty-five patients (23%) had an indeterminate nodule in more than one location.

Table 1.

Frequency of reported findings in 149 patients with staging CT scans of the chest or chest/abdomen/pelvis

Natural History of Indeterminate Nodules

In order of decreasing frequency, indeterminate nodules progressed to clear metastatic disease in four of 13 lymph nodes ≥ 1.0 cm, 31% of lung nodules (15 of 49), one of 15 liver lesions, 4.5% of subcentimeter lymph nodes (two of 44), and zero of four bone lesions (Table 2). All patients with pulmonary incidental findings who developed clear metastatic disease showed progression within 6 months of diagnosis. Nine of 15 lung nodules were determined to be metastatic radiographically, whereas six of 15 were biopsy-proven metastases. Metastatic disease was determined radiographically in all lymph node and liver indeterminate lesions. Of the 35 patients with indeterminate nodules in more than one location, three (9%) had metastatic progression in both locations, whereas 10 (29%) had metastatic disease in one location while the other remained stable.

Table 2.

The ultimate identity of indeterminate nodules in staging CT scans in patients with sarcoma

Factors Associated With Indeterminate Nodules Being Metastatic

A total of 36% of the study group (54 of 149) presented with or developed pulmonary metastases in the study period. Ten of the 54 patients (19%) had evidence of metastatic disease at the time of the initial staging CT scan, 15 patients (28%) had indeterminate nodules, and 29 patients (54%) showed no evidence of metastasis or indeterminate lesions at the time of diagnosis (Table 3). Similarly, four of five with liver metastasis, 11 of 17 with bone metastasis, and five of 13 with lymph node metastasis showed no evidence of metastatic disease or indeterminate nodules at the time of presentation.

Table 3.

Appearance of staging CT scan in patients with metastatic disease

For lung and lymph node incidental findings, primary tumor size ≥ 14 cm was more suggestive of true metastatic disease than for smaller primary tumors (eight of 10 compared with seven of 36 [19%]; odds ratio [OR], 16.6; 95% confidence interval [CI], 2.9-95.9; p < 0.001 for lung; six of 18 compared with zero of 36 [0%], p < 0.001 for lymph nodes) (Tables 4, 5). With the numbers we had, we did not find a clear association with the size of the incidental lung nodule and the likelihood of progression: indeterminate lung lesions 7 to 10 mm were metastatic in four of nine patients compared with 11 of 40 (28%) for those < 7 mm (OR, 2.1; 95% CI, 0.5-9.3; p = 0.427). Similarly, four of 13 lymph nodes ≥ 1.0 cm at staging were eventually determined to be metastatic compared with two of 44 (4.5%) lymph nodes < 1 cm (OR, 9.3; 95% CI, 1.5-59.0; p = 0.020). Five of the six lymph node metastases occurred in thigh tumors, and four of six were in lymph nodes that directly drained the tumor site. High-grade primary tumors had true metastatic disease in 15 of 40 (38%, p = 0.042) incidental findings in the lung but with the numbers we had, there was no association with high-grade tumors and lymph node metastases (six of 43 [14%], p = 0.319). No indeterminate nodule in any location in tumors that was less than high grade progressed to true metastatic disease. Indeterminate lymph nodes in patients with a primary tumor of the thigh were metastatic in five of 20 patients compared with one of 36 (3%) in other locations (OR, 11.7; 95% CI, 1.3-108.6; p = 0.017). We performed the same analysis for bone and liver lesions as was done for lung (Table 4) and lymph nodes (Table 5), but there were no observed associations.

Table 4.

Clinical features and the likelihood that indeterminate lung nodules represent true metastatic disease

Table 5.

Clinical features and the likelihood that indeterminate lymph nodes represent true metastatic disease

Discussion

Nearly all patients diagnosed with sarcoma will have a CT scan performed to search for sites of regional and distant spread. Tumor characteristics such as size, grade, depth, and histology are important factors in choosing an appropriate treatment and providing an accurate prognosis of disease recurrence and overall survival [3, 13, 14, 16, 17, 19]. Accurately predicting whether an indeterminate lesion represents true metastatic disease is important, because metastatic disease is consistently the strongest negative prognostic factor for survival in sarcoma, and a clinician’s assessment of an indeterminate lesion may have implications for decisions regarding biopsies, additional imaging studies, surveillance frequency and modalities, systemic treatment options, and surgical resections and reconstructions. Therefore, information that clarifies the true identity of an indeterminate lesion will help with clinical decision-making, ideally maximizing concern in worrisome lesions and minimizing additional testing and invasive procedures when unwarranted. The main purpose of our study was to report the incidence and eventual outcome of abnormal findings, in particular indeterminate lesions that may or may not represent a site of metastasis, at staging CT scan in patients with sarcoma. In 149 patients, we found that 135 patients (91%) had at least one abnormality reported on staging CT scan, 57 (38%) had an indeterminate lymph node, and 49 (33%) had an indeterminate lung nodule.

Our study does have several limitations that deserve mention. First, our minimum followup interval of 6 months may substantially underestimate the true number of patients with metastatic disease on followup. However, because our goal was to investigate known indeterminate nodules, we believe that a minimum of 6 months is a reasonable timeframe to monitor growth in a defined and visible lesion. Importantly, different areas of the country and world may have variable rates of nonneoplastic pulmonary disease such as fungal, bacteria, or particle disease of the lung (our area in the rural Midwest is thought to be one). In these regions, there may be a higher incidence of indeterminate lung nodules, which would increase the number of incidental findings but lead to a smaller estimate of true metastatic disease. Even if our rates of indeterminate lesions are increased, we believe that the clinical risk factors identified can be generalized to other regions. Finally, our study did not investigate overall survival or make an attempt to recommend appropriate surveillance protocols. Both of these issues are very important and should be the focus of further investigation. Our study is also limited by small numbers resulting in imprecise estimations. Although our data are not robust enough to allow for a true prediction of risk, the findings contribute some information of potential clinical utility for a common clinical concern with limited guidance to date.

In our current study, 91% of patients (135 of 149) had at least one abnormality at staging and 58% (86 of 149) had at least one site with an indeterminate nodule or enlarged lymph node. This percentage of patients is higher than that of a similar study done in patients with prostate cancer [10]. A possible explanation for this is that the prostate study did not include CT scans of the chest, and common abnormal findings in the lungs, chest wall, and thyroid were therefore not seen. Although the primary purpose of staging CT scans is to identify metastatic disease, they routinely detect indeterminate nodules and incidental findings; oftentimes these findings are unimportant and unrelated to the sarcoma and do not change the course of treatment or eventual outcome [14, 16, 20, 21]. The implication is that most incidentally identified lesions represent benign or nonneoplastic processes, and any further imaging, invasive treatment, or patient stress is of no definable benefit to the patient [20, 21].

Our study demonstrated that 36% (54 of 149) of patients either presented with or developed lung metastasis during clinical treatment. Several other studies have conducted similar analyses, documenting a range of 9.6% to 40% [1, 5, 14, 16, 17, 23]. Prior studies have reported a percentage of patients with sarcoma with indeterminate pulmonary nodules ranging from 21% to 67% [17, 20, 21] with 11.5% to 28% eventually proven to be malignant. Our cohort showed similar results because 33% (49 of 149) of our patients presented with indeterminate pulmonary nodules; 31% (15 of 49) of these indeterminate nodules were eventually proven to be metastatic. Viewed from a slightly different perspective, there were 90 patients in our cohort without evidence of lung nodules at baseline. Of these, 29 (32%) developed pulmonary metastatic disease. Interestingly, the same proportion of patients with and without incidental findings ultimately developed lung metastasis, suggesting that the presence of indeterminate nodules is of little importance to the eventual development of disease. We found that 9% of patients presented with or developed lymph node metastasis during clinical treatment. Similar studies have shown lymphatic involvement ranging from 1.75% to 6% [3, 9, 11, 19, 22], but none, to our knowledge, has investigated the incidence or natural history of indeterminate lymph nodes. To this point, we found that 38% (57 of 149) of patients presented with indeterminate lymph nodes with 11% (six of 57) of these nodes eventually proven to be metastatic.

We did find several clinical scenarios that suggest when indeterminate lesions may represent true metastatic disease and may assist clinicians in treatment decisions or patient counseling. First, indeterminate lesions in a primary tumor size ≥ 14 cm progressed to true metastatic disease in eight of 10 lung nodules but with the numbers we had, we could not demonstrate an association with the size of the lymph node and eventual lymph node metastases. With the numbers we had, we could not find evidence for progression to metastatic disease in lung lesions ≥ 7 mm (four of nine) and lymph nodes 1 to 2 cm (four of 13). Prior studies have reported on the effect of nodule size, indicating that patients presenting with pulmonary nodules ≥ 5 mm were at higher risk of metastatic disease than smaller nodules [17, 21]. Finally, we did not find any indeterminate lesions that progressed to clear metastatic disease in low- or intermediate-grade tumors. Taken together, indeterminate lesions in extremely large, high-grade tumors are more likely to represent true metastases, and treatment decisions and patient counseling should acknowledge the potential for progression of disease. Indeterminate lesions in patients with less than high-grade tumors are unlikely to be malignant; therefore, subsequent management should take this into consideration.

In conclusion, it is extremely common for incidental findings and indeterminate nodules to be present at the time of staging CT scan in patients with sarcoma. The need to differentiate between findings that may or may not be clinically important, and if further investigation or intervention is warranted, is a familiar and challenging clinical scenario. We found that many indeterminate nodules are ultimately unrelated to the underlying sarcoma and may not require a change in treatment or surveillance. In patients with large sarcomas, these findings are more likely to be related to the primary malignancy. However, our study is small and limited to a minimum followup of 6 months, and until confirmed in larger studies, we recommend continued monitoring of patients with suspicious findings over time is advised to ensure stability.