Molecular Imaging Biomarkers for Oncology Clinical Trials

Predictive and Prognostic Markers

Predictive markers indicate the aggressiveness of a particular patient’s cancer, the likely course of disease, or the likelihood of response to treatment. Prognostic markers are a class of predictive markers that assess the aggressiveness of a particular cancer and the likelihood it will lead to death. In principle, prognostic measures are intrinsic to the patient and the cancer and are independent of treatment. On the other hand, therapeutic predictive markers describe the likelihood of response to specific treatments and can be used to select treatment most likely to be effective for a particular patient’s cancer. In practice, markers may have both prognostic and predictive features. An example is ER expression in breast cancer. ER expression is both a prognostic and a predictive marker. ER expression predicts a less aggressive tumor and more favorable disease course compared with non–ER-expressing tumors. In addition, a lack of ER expression predicts that breast cancer response to endocrine therapy is unlikely (2). More recently, the use of multigene assays such as Oncotype DX (Genomic Health Inc.) has helped predict tumor behavior and therapeutic response. Such assays are commonly used to determine the need for aggressive treatment such as adjuvant chemotherapy (5).

Early-Response Markers

Early-response markers, sometimes referred to as pharmacodynamic markers, indicate that there has been a biologic response to treatment, namely that the chosen cancer drug has produced an effect on the tumor. The lack of a pharmacodynamic effect indicates that the drug is unlikely to work. Identifying a lack of response, and therefore therapeutic futility, may be the most important use of early-response indicators. An example of a tissue-based early-response indicator is serial assay of tumor proliferation, such as using Ki-67 immunohistochemistry to assess early response to breast cancer endocrine therapy (10).

Surrogate Endpoints

A surrogate endpoint is a marker of therapeutic success that is strongly associated with patient outcomes such as disease-free survival, progression-free survival, or overall survival. An example of a tissue surrogate endpoint is complete pathologic response of breast cancer to presurgical (neoadjuvant) systemic therapy, which has been accepted by the U.S. Food and Drug Administration as a surrogate endpoint for efficacy of breast cancer treatment (11).

Molecular Imaging as a Predictive Marker

Perhaps the clearest example of molecular imaging as a predictive biomarker is PET imaging of ER expression, most widely performed using ¹⁸F-fluoroestradiol. ¹⁸F-fluoroestradiol uptake has been shown to correlate with tissue-based assays of ER expression (12). As with tissue-based ER assays, ¹⁸F-fluoroestradiol PET is predictive of breast cancer endocrine responsiveness, and importantly, a lack of ¹⁸F-fluoroestradiol uptake strongly predicts a lack of response. Early trials on other ER-expressing cancers such as endometrial cancer also support a potential predictive role (13). Through the support of the National Cancer Institute (NCI) Cancer Imaging Program, there is now an NCI-held investigational new drug application (IND) for ¹⁸F-fluoroestradiol to support clinical trials, and there has been successful completion of a small prospective phase 2 trial (Fig. 2) (14). This approach may be helpful for trials of novel endocrine therapies for breast cancer, especially to help select patients whose tumor expresses the therapeutic target (ER); however, more rigorous, carefully controlled, and properly powered studies to establish the negative predictive value of the absence of ¹⁸F-fluoroestradiol uptake with well-defined uptake thresholds are needed to support this indication. Multicenter trials performed under the NCI IND that can address this need are being planned. Several other molecular imaging approaches for measuring regional expression of therapeutic cancer targets are also emerging, including progesterone receptor, HER2, and epidermal growth factor receptor (15).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Images of ER expression using ¹⁸F-fluoroestradiol (FES) PET as example of predictive molecular imaging marker. ¹⁸F-fluoroestradiol and ¹⁸F-FDG PET images are shown from 2 patients, one in whom ¹⁸F-fluoroestradiol confirms ER expression (left) and another in whom imaging demonstrates that recurrent tumor has lost ER expression (right). (Reprinted with permission of (14).)

Molecular Imaging as an Early-Response Indicator

There are several studies showing the potential of ¹⁸F-FDG PET as an early indicator of response to therapy for breast and other cancers. Studies have shown that, on average, a decline in ¹⁸F-FDG uptake early in the course of chemotherapy is predictive of subsequent response and patient survival, although the ability to identify nonresponders is limited by overlap in ¹⁸F-FDG uptake levels between different categories of response in many trials (7,16). Early posttherapy ¹⁸F-FDG PET is increasingly used in therapeutic clinical trials as an early-response indicator, including use as an integral biomarker to direct therapy selection in ongoing lymphoma trials (7). ¹⁸F-FDG has also been shown to be predictive of breast cancer response to endocrine therapy, indicated by either an increase in ¹⁸F-FDG uptake (a flare in response to estrogen agonists) (17) or a decline in early response to antagonists (18). Recent results from the NeoALTTO trial showed a similar early predictive value for HER2-directed therapy (9). These results support the increasing use of ¹⁸F-FDG PET as both an integrated and an integral marker in trials of targeted therapy; however, well-designed prospective studies are needed to better establish this role for ¹⁸F-FDG PET.

Molecular imaging probes beyond ¹⁸F-FDG may be even more sensitive and robust indicators of early drug response. Following the analogy of tissue-based assay of cellular proliferation as early-response indicators, trials of ¹⁸F-fluorothymidine PET have shown its value in assessing response as early as 1 wk after initial treatment (Fig. 3) (19). ¹⁸F-fluorothymidine PET is being tested as an early-response indicator in multicenter trials, including ACRIN 6688, a trial of early breast cancer response to neoadjuvant chemotherapy performed by the American College of Radiology Imaging Network (ACRIN) under an NCI IND for ¹⁸F-fluorothymidine (20). This trial has completed accrual, and results are expected within the next year.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

¹⁸F-fluorothymidine PET/CT studies at baseline (A), within 1 wk of completion of first cycle of neoadjuvant chemotherapy (B), and after completion of 6 cycles of neoadjuvant chemotherapy and before mastectomy (C). Baseline images show biopsy-proven multifocal disease in right breast on both PET and CT (arrows and arrowheads). After single cycle of chemotherapy, uptake has qualitatively decreased on PET but remains on CT (arrowhead), whereas residual nodule remains on both PET and CT (arrows). After completion of neoadjuvant therapy, only residual nodule remains, though it is less intense. Pathology revealed residual 1.7-cm focus of invasive carcinoma with complete response elsewhere in breast.

Molecular Imaging as a Surrogate Endpoint

Imaging is frequently used as a surrogate endpoint for therapeutic trials, most commonly based on changes in tumor size assessed from anatomic imaging according to standard criteria such as the Response Evaluation Criteria in Solid Tumors (RECIST) (21). Although size-based surrogate response endpoints have had success and are commonly used in clinical trials, they have several limitations, including the inability to distinguish residual tumor from nontumor residual tissue masses and significant challenges for assessing response to cytostatic agents when tumor cell death and reduction in tumor mass is not expected. Molecular imaging is beginning to be used in surrogate endpoint criteria—for example, as an adjunct for identifying disease progression in RECIST 1.1—though it is not widely accepted as a surrogate endpoint for treatment outcome by regulatory agencies. This fact motivates ongoing efforts to establish molecular imaging, especially ¹⁸F-FDG PET, as an accepted response endpoint for cancer therapy trials. These efforts have been most successful in lymphoma, for which numerous studies have supported the value of negative posttherapy PET results in predicting survival, and for which ¹⁸F-FDG PET is commonly used as an endpoint in clinical trials (22). Although single-center trials have supported the ability of ¹⁸F-FDG PET to predict survival in other tumors, including breast cancer (23,24), further prospective studies are needed to support more widespread acceptance of molecular imaging as a surrogate response endpoint. Molecular imaging may be especially helpful in some clinical settings, such as bone metastasis response, in which anatomic imaging is ineffective and early data support the effectiveness of PET as a response endpoint (25).