Gene Profiling Directs Site-specific Therapy for Carcinoma of Unknown Primary Site

Get Permission

We feel that the body of evidence is sufficient to support the use of molecular tumor profiling in the standard management of patients with [carcinoma of unknown primary].

Patients with carcinoma of unknown primary site usually receive empiric therapy (eg, with taxane/platinum or gemcitabine/platinum regimens), resulting in a median overall survival of approximately 9 months. As reported recently in Journal of Clinical Oncology, Hainsworth and colleagues have shown that molecular tumor profiling with a 92-gene assay can predict tissue of origin in most patients, and that assay-directed site-specific treatment is associated with overall survival that compares favorably with that seen with empiric therapy.

Study Details

The study prospectively enrolled 289 patients with carcinoma of unknown primary. Patients had to have Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2, no prior systemic chemotherapy, measurable or evaluable disease, and adequate organ function. Patients were excluded if they had a single resectable metastasis.

Of the 289 patients, 252 (87%) had successful molecular profiling assays, with tumor samples or quantities of viable RNA in samples being inadequate for the remainder. Among the 252 patients with successful assays, median age was 64 years (range, 26–89 years), 54% were female, and 69% had two or more metastatic sites. Histology consisted of adenocarcinoma in 51%, poorly differentiated adenocarcinoma in 24%, poorly differentiated carcinoma in 18%, squamous carcinoma in 5%, poorly differentiated neuroendocrine carcinoma in 1%, and poorly differentiated neoplasm in 1%.

Four Tissues Account for 55% of Predicted Sites

The assay predicted a tissue of origin in 247 (98%) of the 252 patients. Twenty-six different tissues of origin were identified. These included the biliary tract in 21%, urothelium in 12%, and colorectum and non-small cell lung in 11% each, with these sites accounting for 55% of all predicted sites. Other predicted tissues of origin included: pancreas and breast in 5% each; ovary, gastroesophageal, and kidney in 4% each; and liver in 3%. Of the assay predictions, 119 (48%) were made with 80% or greater probability.

Of the 252 patients with successful assays, 223 (88%) remained on study and received treatment. Of the 29 patients not remaining on study, 16 no longer met eligibility criteria (declining performance status in 15, brain metastases in 1) and 13 discontinued the study as a result of patient decision (refusal of recommended treatment) or physician preference (usually doubt regarding assay findings). Of these 223 patients, 194 (87%) received assay-directed therapy; of the 29 patients not receiving assay-directed therapy, 23 received treatment preferred by their physician, 5 received protocol-directed therapy for unclassifiable tumors, and 1 had rapid deterioration of condition during palliative radiotherapy.

Site-specific therapies consisted of the following: taxane/bevacizumab (Avastin) for breast; FOLFOX (leucovorin, fluorouracil [5-FU], and oxaliplatin) or FOLFIRI (leucovorin, 5-FU, and irinotecan), or variants, plus bevacizumab for colorectal; platinum-based doublet plus bevacizumab for non–small cell lung; paclitaxel/carboplatin plus bevacizumab for ovary; gemcitabine/erlotinib (Tarceva) for pancreas; sunitinib (Sutent) or bevacizumab with or without interferon for renal; and standard guideline-based first-line treatment for other diagnoses.

Survival Findings

Among the 194 patients receiving assay-directed site-specific therapy, median overall survival was 12.5 months (95% confidence interval [CI] = 9.1–15.4 months). Median overall survival was 13.4 months in 115 patients with tumor types considered to be treatment-responsive (ie, colorectal, breast, ovary, kidney, prostate, bladder, non–small cell lung, germ cell, poorly differentiated neuroendocrine, and small cell lung cancers) and 7.6 months (P = .04) in 79 patients with less-responsive tumor types (ie, biliary tract, pancreas, gastroesophageal, liver, sarcoma, cervix, carcinoid, endometrium, mesothelioma, melanoma, skin, thyroid, head and neck, and adrenal cancers).

Analysis by prediction probability and tumor responsiveness showed median overall survival of 15.4 months for the treatment responsive/≥ 80% probability subgroup (n = 42), 13.4 months for the treatment responsive/< 80% probability subgroup (n = 57), 8.2 months for the less-responsive/≥80% probability subgroup (n = 37), and 7.0 months for the less responsive/< 80% probability subgroup (n = 58), with P = .03 for the trend. For the most common predicted tumor types, median overall survival was not reached at more than 24 months for breast cancer (n = 10), while for others was reached at 29.6 months for ovary (n = 10), 15.9 months for non–small cell lung (n = 23), 12.5 months for colorectal (n = 26), 11.7 months for renal (n = 9), 8.4 months for urothelium (n = 27), 8.2 months for pancreas (n = 12), and 6.8 months for biliary tract cancers (n = 45).

Need for More Evidence?

The investigators posed the question of how much additional evidence would be necessary for molecular profiling to be accepted as a standard component of diagnosis for patients with carcinoma of unknown primary. A randomized phase III trial would provide definitive evidence, but such a trial would have inherent difficulties and would require large numbers of patients in order to generate conclusions regarding outcomes for the numerous potential tissues of origin.

Moreover, interpretation of such a study would be complicated by the fact that first-line empiric therapy is identical or similar to standard site-specific therapy for some of the more-responsive tumor types and that patients who might be expected to benefit from site-specific vs empiric therapy would constitute a minority of the study population. In addition, many patients and physicians may be reluctant to accept randomized assignment to treatment in such a setting.

As stated by the investigators, “At this time, we feel that the body of evidence is sufficient to support the use of molecular tumor profiling in the standard management of patients with [carcinoma of unknown primary]. Patients who will benefit most … are those predicted to have treatment-responsive tumor types. With an increasing number of effective molecularly targeted therapies available, an assay prediction may also lead to the identification of additional therapeutic options. For example, an assay prediction of non–small cell lung cancer could lead to the identification of unsuspected EGFR or ALK mutations. A sizable percentage of patients with [carcinoma of unknown primary] will not currently benefit from assay-directed therapy, because effective therapy for these tumor types does not currently exist. However, confidence in the assay predictions will allow these patients to receive effective therapy as the standard therapy for these tumor types improves.” ■


1. Hainsworth JD, Rubin MS, Spigel DR, et al: Molecular gene expression profiling to predict the tissue of origin and direct site-specific therapy in patients with carcinoma of unknown primary site: A prospective trial of the Sarah Cannon Research Institute. J Clin Oncol. October 1, 2012 (early release online).