Use of 64Cu-DOTA-Trastuzumab PET to Predict Response and Outcome of Patients Receiving Trastuzumab Emtansine for Metastatic Breast Cancer: A Pilot Study

We hypothesized that functional imaging with 64Cu-DOTA-trastuzumab PET/CT would predict the response to the antibody–drug conjugate trastuzumab–emtansine (T-DM1). Methods: Ten women with metastatic human epidermal growth factor receptor 2–positive breast cancer underwent 18F-FDG PET/CT and 64Cu-DOTA-trastuzumab PET/CT on days 1 and 2 before treatment with T-DM1. Results: T-DM1–responsive patients had higher uptake than nonresponsive patients. Day 1 minimum SUVmax (5.6 vs. 2.8, P < 0.02), day 2 minimum SUVmax (8.1 vs. 3.2, P < 0.01), and day 2 average SUVmax (8.5 vs. 5.4, P < 0.05) for 64Cu-DOTA-trastuzumab all favored responding patients. Tumor-level response suggested threshold dependence on SUVmax. Patients with a day 2 minimum SUVmax above versus below the threshold had a median time to treatment failure of 28 mo versus 2 mo (P < 0.02). Conclusion: Measurement of trastuzumab uptake in tumors via PET/CT is promising for identifying patients with metastatic breast cancer who will benefit from T-DM1.

We have used 64 Cu-DOTA-trastuzumab PET/CT to study women with recurrent or metastatic breast cancer (4) and reported a positive correlation between tumor uptake of 64 Cu-DOTA-trastuzumab and HER2 status as measured by immunohistochemistry (4,5). Trastuzumab-emtansine (T-DM1) is an antibody-drug conjugate that uses trastuzumab to target HER2-positive breast cancer and deliver its cytotoxic payload, emtansine (6). T-DM1's mechanism of action and use as a single agent are advantageous for correlating trastuzumab imaging with treatment response. We report the results of a pilot study testing whether pretreatment 64 Cu-DOTA-trastuzumab PET/ CT can predict benefit from T-DM1 for HER2-positive metastatic breast cancer.

Patient Selection
Eligibility requirements included metastatic or recurrent HER2-positive breast cancer in patients who were to receive T-DM1, were 18 y old or older, had an Eastern Cooperative Oncology Group performance status of 0-2, had a normal cardiac ejection fraction, and had at least 1 metastasis with a diameter of at least 2.0 cm. Patients could not have received trastuzumab for 4 or more weeks. Eligibility was confirmed by tumor biopsy for HER2 assessment and 18 F-FDG PET/CT. The City of Hope Institutional Review Board approved the study, and all patients provided written informed consent (NCT02226276).

Treatment
The patients underwent a clinical examination and toxicity assessment before each cycle of T-DM1. Follow-up 18 F-FDG PET/CT was performed after every 2 cycles of T-DM1 for up to 18 mo and at the discretion of the treating oncologist thereafter. Treatment response was evaluated by PERCIST (7). 64 Cu-DOTA-Trastuzumab-PET/CT 64 Cu was provided by the Mallinckrodt Institute of Radiology, Washington University School of Medicine. Radiolabeled trastuzumab was prepared and administered according to IND 109971 as previously described (4).
Scans were acquired with a Discovery STe 16 PET/CT device (GE Healthcare) operated in 3-dimensional mode. The PET axial field of view and slice thickness were 15.4 cm and 3.3 mm, respectively. PET images were iteratively reconstructed as previously described (5) and had a measured resolution of 9 mm in full width at half maximum. 64 Cu-DOTA-trastuzumab PET/CT was performed during the first day (day 1) and second day (day 2) after injection. Quantitative imaging with 64 Cu-DOTA-trastuzumab was supported by direct measurement of activity concentrations in peripheral venous blood samples drawn before imaging on days 1 and 2. Measurement of 64 Cu-DOTA trastuzumab uptake is described in Supplemental Figure 1 (supplemental materials are available at http://jnm.snmjournals.org).
Antibody scans were interpreted in relation to baseline 18 F-FDG scans by a radiologist different from those who evaluated the 18 F-FDG PET/CT scans. Quantitative analysis was performed using XD (version 3.6; Mirada Medical). Correction for altered 18 F-FDG biodistribution in follow-up examinations is shown in Supplemental Figure 2.
Tumor uptake was measured in terms of voxel SUV max (4,5). Measurements were limited to tumors measurable for baseline 18 F-FDG uptake. Tumor images strongly overlapped by positively imaged adjacent features (e.g., a vessel or organ) were rejected for uptake measurement. Those that were strongly positive and well separated from adjacent features were segmented via a maximum voxel-based thresholding technique (8). The methods used for tumor assessment and numbers of tumors assessed are in Supplemental Tables 1 and 2.

Statistical Plan and Analysis
The study was designed to accrue 10 patients to explore the relationship between tumor uptake of 64 Cu-DOTA-trastuzumab and tumor response. For individual tumors, a hierarchic (tumor-within-patient) linear mixed-effects model was used to evaluate the association between day 1 or day 2 SUV max and response. The best SUV max cut points for individual tumors were used in patient-level response (Fisher exact test), and planned comparisons of average (or minimum) uptake in responsive versus nonresponsive patients used the t test. All P values are 2-sided. Cox regression was used for time to treatment failure (TTF) (the supplemental materials provide additional details).

RESULTS
Ten patients were enrolled, and their characteristics are in Supplemental Table 3. Five experienced a response, and 5 were nonresponders; TTF ranged from 1.3 mo (early death) to 46 mo. Two patients continued in long-term follow-up at 62 and 78 mo from the initiation of chemotherapy. Fifty-nine 18 F-FDG baseline-measurable tumors met the criteria for measurability of 64 Cu-DOTA-trastuzumab uptake, and 31 (day 1) and 25 (day 2) were also measured for response. Over half the data for individual tumors came from 2 patients (Figs. 1B and 1C).
Individual tumor response appeared to have a distinct threshold dependence on uptake of 64 Cu-DOTA-trastuzumab, especially on day 2 (Fig. 2). Although the optimal uptake threshold settings accurately separated responsive from nonresponsive tumors, those results are not statistically significant in a hierarchical model of tumors within patient for this 10-patient cohort.
Patient-level response was positively related to tumor uptake of 64 Cu-DOTA-trastuzumab, and the thresholds that optimally related uptake to best response for individual tumors also accurately separated patients by response to T-DM1 (Fig. 3). Responsive patients had a significantly higher day 2 average, day 1 minimum, and day 2 minimum SUV max than nonresponsive patients. In the categoric analysis (intrapatient average or minimum tumor SUV max . response threshold, yes/no, vs. responsive, yes/no), day 1 results were significant for minimum SUV max , whereas day 2 results were significant for both metrics. For day 2, all patients with the lowest tumor uptake above the threshold responded, whereas no patients with the lowest tumor uptake below the threshold responded.
TTF was positively related to measured tumor uptake of 64 Cu-DOTA-trastuzumab (Table 1). For day 2, the relationship was statistically significant for both patient-level uptake metrics. Depending on the metric, the day 2 tumor response threshold discriminated patients with a median TTF of 2 versus 23 or 28 mo.
The supplemental materials include additional details about patients, treatment, response assessment, and 64 Cu-DOTA-trastuzumab image analysis.

DISCUSSION
We demonstrated a significant association between tumor uptake of 64 Cu-DOTA-trastuzumab and patient benefit (response and TTF) from treatment with T-DM1. The ZEPHIR trial found pretreatment tumor imaging with 89 Zr-trastuzumab PET/CT to be predictive of patient response and TTF in T-DM1 therapy for HER2-positive metastatic breast cancer (9,10). The pretreatment work-up included 18 F-FDG PET/CT. 89 Zr-trastuzumab PET/CT scans acquired 4 d after injection were assessed by radiologists' qualitative inspection. Our quantitative study corroborates the ZEPHIR trial's finding that tumor uptake of trastuzumab on PET/CT correlates with patient response and outcome with T-DM1. Using SUV max measurement, moreover, we found an apparent threshold relationship between tumor response and tumor uptake 1 and 2 d after injection of 64 Cu-DOTA-trastuzumab. The response thresholds for individual tumors also accurately separated patients by response and TTF. Based on published data for 89 Zr-trastuzumab (11,12), we estimate that the 64 Cu-DOTA-trastuzumab SUV max T-DM1 response threshold is modestly greater than blood pool SUV at 4 d postinjection (Supplemental Fig. 3 and associated text). This is consistent with the criterion for "relevant" tumor uptake of 89 Zr-trastuzumab visa-vis response to T-DM1 in the ZEPHIR trial.
The current study demonstrated some disadvantages of 64 Cu (half-life, 12.8 h) relative to 89 Zr (half-life, 3.3 d). 64 Cu does not provide whole-body coverage with an acceptable scan duration and adequate count density for accurate and precise measurement of tumor uptake. In the current study, the low count rate limited coverage of disease burden on day 2. Overlap with images of adjacent blood vessels reduced the number of tumor images measurable for 64 Cu-DOTA-trastuzumab SUV max . Variations in time between injection and scan (t scan 2 t inj ) imposed by difficulty in scheduling research scans amid clinical operations added noise to measurements of tumor SUV max . The resulting error is inversely related to t scan 2 t inj and thus inherently much worse for scans on days 1 and 2 than scans on day 4 or later. The problem is well illustrated by the patient (Fig. 1A) with the shortest day 1 time to scan (16 h). Although the patient had a complete response to T-DM1, SUV max was below the empiric response threshold for both of her measurable tumors on day 1, whereas on day 2, SUV max was above the threshold.
Despite limitations imposed by the relatively short half-life of 64 Cu, we have demonstrated that measurement of tumor uptake of trastuzumab at 1-2 d after injection can be effective in identifying patients unlikely to benefit from T-DM1 therapy. Further work is required to develop measurement of trastuzumab uptake as a predictor of clinical benefit from T-DM1. Trastuzumab imaging may identify patients who could benefit from T-DM1 and thus avoid the toxicity of chemotherapy. Trastuzumab imaging may also identify women who might not otherwise be considered for HER2-directed treatment (13).
We previously reported on the patient depicted in Figure 1C, whose disease demonstrated heterogeneous uptake of 64 Cu-DOTAtrastuzumab and a correspondingly mixed response to T-DM1 (14). This case suggests that trastuzumab imaging may identify patients who could benefit from combining T-DM1 with chemotherapy or other treatments.

CONCLUSION
Tumor uptake of 64 Cu-DOTA-trastuzumab, measured in terms of SUV max at 1-2 d after injection, was positively associated with patient response and TTF in T-DM1 therapy of HER2-positive metastatic breast cancer. The relationship between SUV max and tumor response appeared to have a sharp threshold, and the threshold for individual tumor response was also effective in separating patients who did and did not benefit from T-DM1. Thus, measurement of trastuzumab uptake in tumors via PET/CT is highly promising for patient selection in treatment of metastatic breast cancer with T-DM1.  . Relationship between patient best response to T-DM1 and measured tumor uptake of 64 Cu-DOTA-trastuzumab. Dashed lines show optimal thresholds relating response to uptake for individual tumors (day 1, 4.6 g/mL; day 2, 5.5 g/mL). Group mean SUV max (responsive vs. nonresponsive patients) was significantly different (t test) for day 2 average (8.5 vs. 5.4, P , 0.05), day 1 minimum (5.6 vs. 2.8, P , 0.02), and day 2 minimum (8.1 vs. 3.2, P , 0.01). (P 5 0.08 for day 1 average). *Fisher exact test for response vs. threshold.