We invited 316 physicians to participate in this study. Of these, 152 did not respond to recruitment efforts, 82 declined participation, and 9 consented to participate but did not finalize data entry. Eight physician reports failed to meet inclusion criteria. The final sample consisted of 65 physicians who had ordered and received a GEP test for 107 patients between July 2009 and December 2009.
Most physicians in the study (62%) were part of a group practice and 57 (88%) were board-certified in medical oncology (Table ). Sixty percent reported seeing at least six patients per year with difficult-to-diagnose tumors. Sixty-one (57%) patients were women; mean patient age was 64 (standard deviation [SD] 12), and 54 (50%) patients were at least 65 years old. Of the 91 patients with a reported Eastern Cooperative Oncology Group performance status score, 87 (81%) had a score of 2 or less.
Physician and patient characteristics
The tissue samples submitted for GEP testing came from 19 biopsy sites. The most common sites were lymph node (21%), soft tissue (20%), liver (18%), lung (9%), bone (7%), and brain (5%). Participating physicians ordered 343 imaging or endoscopic investigations, for an average of 3.2 investigations per patient (Table ). All patients had an imaging study; all but 1 patient had imaging with a magnetic resonance imaging, computed tomography, or positron emission tomography. That patient had ovarian cancer, which was imaged with ultrasound. The most common test, ordered for 85% of patients, was a computed tomography scan. Overall, imaging and endoscopic tests detected a tumor in more than 19 sites, with the most common sites being lymph node, liver, and lung. Physicians ordered 147 IHC tests, with a mean of 9.9 (SD 5.2) IHCs per biopsy.
Pathology and imaging or endoscopic investigations
Before receiving the GEP test results, physicians reported having no working diagnosis for 41% (44) of patients (Figure ). The most common pretest tumor-site working diagnoses reported for the other 59% (63) of patients were lung (9), cholangiocarcinoma (7), breast (6), colon (5), ovarian (5), gastric (4), and pancreas (3).
Proportion of site-specific diagnoses before and after GEP
After receiving the GEP test results, physicians changed the primary working diagnosis site for 54 patients (50%, 95% CI: 43%, 58%; p<0.0001). The number of patients with no working diagnosis declined from 44 to 17 (25% difference, 95% CI: 15%, 36%; p<0.0001; Table ). The tissue-site working diagnosis remained unchanged in 37 (35%) cases. For example, of the 10 cases that specified lung as the initial tissue-site working diagnosis, 7 remained specified as lung after GEP test; the tissue-site working diagnosis similarly remained unchanged before and after GEP testing in 3 of 7 cholangiocarinomas, 5 of 6 breast cancers, 2 of 5 colon cancers, 5 of 5 ovarian cancers, and 1 of 3 pancreatic cancers. More cancers had a reported site working diagnosis of colon, lung, pancreas, or breast after GEP. In contrast, the number of patients with a cholangiocarcinoma working diagnosis decreased from seven to three after the test, and the number of gastric cancer working diagnoses dropped from four to two. The most common GEP results were colorectal (19%), breast (14%), sarcoma (14%), lung (11%), ovarian (11%), and pancreas (10%) tissue types. The most common working diagnoses after GEP were lung (13%), colorectal (11%), pancreas (9%), and ovarian (8%) cancer.
Tissue type working diagnoses before and after GEP results
Overall, for 70 patients (65%, 95% CI: 58%, 73%; p<0.0001) physicians changed one or more management recommendation after GEP (Table ). For 17 patients, physicians did not change the primary working diagnosis but changed management recommendations. Prior to the test, physicians reported recommending 44 different chemotherapy drug combinations for 75 patients and no chemotherapy for 32 patients. Carboplatin plus paclitaxel was recommended in 26 patients. Physicians changed their recommended chemotherapy regimens after receiving GEP results for 58 patients (54%, 95% CI: 46%, 62%; p<0.0001, Table ). With GEP results, the recommendation for chemotherapy regimens that are consistent with guidelines for the final tissue-type working diagnosis increased by 23%, from 42% to 65% of patients, and the recommendation for regimens not consistent with guidelines for the final tissue-type working diagnosis declined by 15%, from 28% to 13%.
Median survival from time of the biopsy was 14.0 months (95% CI: 10.2, 18.6), with 33% of patients alive at 2 years (95% CI: 24%, 44%) and 30% of patients alive at 3 years (95% CI: 20%, 41%; Figure ). Median time from biopsy to requisition of GEP was 1.0 months (95% CI: 0.9, 1.3). The most common final working diagnoses for patients alive at 3 years were breast (7), lung (7), colorectal (5), ovarian (4), and no working diagnosis (4). Prior to GEP test results, for the 61% of the patients for whom physicians had identified a pretest primary site, physicians ranked conventional pathology (including IHC testing) as the most influential factor for diagnosis in 53% of patients. After receiving the GEP results, for these patients, physicians cited the GEP test as the most influential factor for diagnosis in 60% of the patients; physicians cited conventional pathology as the most influential factor for diagnosis in 23% of all patients.
When asked whether the test changed their approach to treatment, physicians reported agreeing or strongly agreeing for 52% of patients. Two-thirds of physicians agreed or strongly agreed that the test results were clinically useful in the care of their patient. Compared with leaving the tissue-site working diagnosis unspecified after GEP testing, being able to specify a tissue-site working diagnosis or not changing a previously-specified tissue site was associated with agreement or strong agreement that the test was clinically useful. Changes in surgical plans, orders for more investigations, or both were associated with agreement that the GEP test was clinically useful.
Physicians stated that the GEP test was helpful because it (1) provided guidance in choosing more targeted therapy, (2) confirmed a working diagnosis, (3), helped alleviate patient anxiety, (4) led to a more appropriate further workup or triaging to a more appropriate specialist, or (5) helped rule out the least likely possible diagnoses. Of physicians who did not agree that the GEP test was helpful, 10 (9%) reported that the likely diagnosis was not on the report panel and 6 (6%) did not believe the results.