Copper-64 trastuzumab PET imaging: A reproducibility study

Abstract

BACKGROUND:

The current study aims to assess the safety, pharmacokinetics, feasibility, and reproducibility of immunoPET imaging with copper-64 (⁶⁴Cu)-trastuzumab.

METHODS:

An i.v. injection of 296–370MBq/5mg ⁶⁴Cu-trastuzumab was administered between 1 to 4 h after routine trastuzumab treatment. Whole-body PET scans were performed immediately post-injection and at 24 h post-injection. Serial pharmacokinetics were performed. Of 11 patients (median age of 52; range of 31–61), 8 underwent a repeat study with ⁶⁴Cu-trastuzumab to assess image and pharmacokinetic reproducibility. Patients were monitored for toxicity.

RESULTS:

Patients experienced no allergic reactions or significant adverse effects from ⁶⁴Cu-trastuzumab. Eight patients successfully completed a repeat ⁶⁴Cu-trastuzumab study, with acceptable reproducibility of both the biodistribution and pharmacokinetic clearance. Study 1 vs. study 2 showed similar serum concentration post-injection (mean 42.4 ± 7.8 %ID/L vs. 44.7 ± 12.6 %ID/L) and similar T1/2 (single exponential 46.1 vs. 44.2 h), p>0.5. The volume of distribution (median 2.50L) was in the range reported for trastuzumab and close to the estimated plasma volume of 2.60L. Of 11 patients, 2 had ⁶⁴Cu-trastuzumab localization corresponding to known tumor sites—1 in liver and 1 in breast.

CONCLUSIONS:

Preliminary results suggest that scanning with ⁶⁴Cu-trastuzumab is feasible, safe, and reproducible. Tumor uptake of ⁶⁴Cu-trastuzumab was observed, but tumor detection exhibited low sensitivity in this study in which imaging was performed in the presence of trastuzumab therapy.

INTRODUCTION

Human epidermal growth factor receptor-2 (HER2) is a transmembrane receptor in the HER family of receptor tyrosine kinases that is over-expressed in up to 25% of breast cancer and to a lesser extent in other malignancies. In breast cancer it is associated with more aggressive disease and poorer prognosis (1). Because of the role of HER2 in the pathogenesis of HER2 breast cancer, it has been a target for therapeutic intervention with both antibodies, such as trastuzumab, and small molecules (2). Although survival benefit in patients receiving trastuzumab has been shown, patients often develop relative treatment resistance, and thus new approaches to target this pathway are being actively investigated (2).

Imaging with radiolabeled trastuzumab may allow for monitoring expression of HER2 status prior to and during therapy. Various groups have reported on the heterogeneity of tumor HER2 expression (3–6). Furthermore, it has been published that in the adjuvant setting, patients classified as HER2-negative sometimes benefit from HER2-directed therapy, possibly due to the presence of HER2-positive tumor cells that were not biopsied directly (7) or variabilities in the assay or its interpretation. Other reports have shown a lack of efficacy in patients with HER2-negative biopsied disease (8). If non-invasive imaging with trastuzumab can accurately assess this heterogeneity, it may be possible to stratify or select patients for intervention based on HER2 positivity of tumors, optimize therapy in individual patients, or monitor target modulation.

Several studies have utilized radiolabeled trastuzumab for radioimmunodetection of HER2 in the preclinical or clinical setting (9–11). The use of intact IgG is associated with slow targeting kinetics and thus requires long-lived isotopes to adequately visualize lesions. Given the higher sensitivity, resolution, and quantitative capability of PET compared to planar or SPECT imaging, interest in the use of long-lived positron emitters suited to the slow kinetics of intact IgG is increasing (12–17). However, the long half-lives of ¹²⁴I and ⁸⁹Zr (100.2 h and 78.4 h, respectively) are sub-optimal for the purpose of minimizing dose burden and duration between repetitive short-term pharmacodynamic PET studies. This led to our interest in utilizing shorter-lived positron emitters such as ⁶⁸Ga and ⁶⁴Cu for antibodies or antibody-like reagents (18–21).

Recent work has reported successful imaging of breast cancer using ⁶⁴Cu-trastuzumab (19, 20), which has a 12.7-hour half-life. The use of ⁶⁴Cu may enable sequential short-term, non-invasive quantification of HER2 expression by PET imaging. Here we report a study with the goal of evaluating the repeatability and reproducibility of the technique.

MATERIALS AND METHODS

This was a prospective, open label imaging protocol performed to evaluate the feasibility of imaging patients with ⁶⁴Cu-trastuzumab; assess its safety, pharmacokinetics, and biodistribution; and evaluate reproducibility of repeat administrations. The protocol was approved by the Memorial Sloan Kettering Institutional Review Board; all patients gave written and oral informed consent. The clinical trial registration number is NCT00605397.

The median age of patients was 52 years (range 31–61 years) (Table 1). All patients were female with biopsy-proven metastatic breast cancer known to express HER2 based on previously documented 3+ immunohistochemistry (n=5) or with gene amplification (>2.0) by fluorescence in situ hybridization (FISH) (n=6) (Table 1). All patients had measurable or evaluable disease and were on a stable trastuzumab therapy regimen either alone (n=1) or combined with chemotherapy (n=10). Eight patients (Table 1) were administered a second ⁶⁴Cu-DOTA trastuzumab to evaluate the reproducibility of repeat imaging.

Table 1.

Patient Characteristics.

Patient	Age	HER2 status	Prior duration of trastuzumab (months)	Trastuzumab dosing schedule / total mass	Time between two studies (days)
1	58	FISH (5.5)	17	2mg/kg (145mg)	28
2	38	3+ by IHC	24	6mg/kg (462mg)	21
3	50	FISH (4.3 & 7.3)	1	2mg/kg (145mg)	Refused
4	31	3+ by IHC	1	6mg/kg (561mg)	42
5	55	FISH (3.5)	12	2mg/kg (240mg)	28
6	32	FISH (3.7)	8	2mg/kg (160mg)	56
7	44	3+ by IHC	40	2mg/kg (100mg)	Unwell
8	52	FISH (4.5)	24	2mg/kg (140mg)	Refused
9	61	3+ by IHC	7	6mg/kg (500mg)	43
10	60	3+ IHC	38	2mg/kg (130mg)	40
11	52	Fish (5.8)	12	6mg/kg (300 mg)	63
Median	52	-	12	2mg/kg (262 mg)	41

Copper-64 trastuzumab

Copper-64 was obtained from Washington University Medical School, where it was produced by proton irradiation of enriched Nickel-64 (22). Clinical grade trastuzumab was obtained (Herceptin™, Genentech, South San Francisco, CA) and reconstituted according to the manufacturer’s instructions and conjugated with DOTA (1,4,7,10-tetraazacyclododecane-N,N’,N’’,N’’’-tetraacetic acid). In brief, the DOTA-NHS was reacted with trastuzumab and the DOTA-trastuzumab conjugate was buffer-exchanged and sterile filtered into vials at 10 mg/mL in 1M acetate buffer. The product was sterile and contained <2 EU/ml, with a mean of 5.3 DOTA chelates per antibody (data not shown).

For labeling, 5 mg of DOTA-trastuzumab was incubated with ~555 MBq of ⁶⁴Cu for 15 min. Unchelated and non-specifically bound ⁶⁴Cu was scavenged by adding 5 mM DTPA, and the entire mixture was then purified through a P6 column. The final product was sterile filtered and formulated in 5% NaCl 1% HSA with 370 MBq nominal activity and 5 mg of trastuzumab. Radiochemical purity was determined using instant thin layer chromatography and was >90%, with each dose containing approximately 3.5 to 4 mg of antibody. Immunoreactivity was estimated using the Lindmo method (23) with HER2-positive BT474 breast cancer cell line. The mean immunoreactivity was 93.5 ± 5.2 (n=19). There was no difference in the immunoreactivity of products for diagnostic study 1 or 2 (n=8, p=0.944).

On the day of ⁶⁴Cu-trastuzumab injection, patients received their current trastuzumab therapy regimen consisting of 2 or 6 mg/kg typically administered over 30 min, before being injected with ⁶⁴Cu-trastuzumab at a median of 98 min (range of 52–173 min) post-injection of cold trastuzumab. The mean activity of ⁶⁴Cu-trastuzumab was 336.7 ± 22.2 MBq (range 296–370 MBq), injected over ~1 min. Eight patients had a repeat study after a median of 41 days (range 31–63). The second study was performed while on the same therapeutic regimen of trastuzumab ± chemo. No significant differences in elapsed time between “cold” and ⁶⁴Cu-trastuzumab were seen between study 1 and study 2 (mean 124 ± 42 vs. 110 ± 38 min), p=ns.

Imaging

Imaging was performed with a GE Discovery STE PET/CT scanner (GE, Waukesha, Wisconsin) in 2D mode with attenuation, scatter, and other standard corrections applied. Images were acquired from mid-skull to proximal thighs on the day of injection (~60 min post-tracer administration) and the day after. On the day of injection, images were acquired at 6 min per field of view and next-day images were acquired at 10 min per field of view.

An experienced nuclear medicine physician with knowledge of the patient’s history and the diagnostic CT scan (JAC) read the images on a dedicated PET analysis workstation (Hermes hybrid viewer, Hermes, Stockholm, Sweden). Localization in tumor was defined as focal accumulation greater than adjacent background in areas where physiologic activity was not expected. Volumes of interest (VOI) were placed visually over structures of interest and mean and maximum SUV normalized to body weight or lean body mass [(nCi/mL activity in region)/(nCi injected activity / body mass or lean body mass in grams)] were determined for blood pool, lung, liver, spleen, kidney, and marrow. In order to determine the %ID/g from the PET image, the SUV was divided by the body weight (in grams) and multiplied by 100. Studies were considered reproducible if SUVs were within 20% of each other; although this is an arbitrary value, with other tracers (such as FDG), changes greater than 20% were considered significant.

Pharmacokinetics

Pharmacokinetic analysis was performed in all patients by drawing blood at ~5 min, 10 min, 30 min, 1 h, 2 h, and 24 h after the injection. All samples were centrifuged and serum was aliquoted, weighed, and counted in a scintillation well counter (PerkinElmer Wizard 2480, Waltham, MA). A standard dose was also counted and the decay-corrected %ID/L in serum was calculated, which served as the gold standard for serum/plasma concentration. The data was fit to a monoexponential rather than bi-exponential function due to limited samples (GraphPad Prism, La Jolla, CA). The fitted half-time and time-zero intercept were used to determine the biologic T_1/2, volume of distribution of central compartment (V_c), area underneath the curve (AUC), and systemic clearance. The AUC was determined by trapezoidal integration up to the time of the last blood samples. In order to estimate the %ID in the plasma volume, the %ID/L at the end of infusion was multiplied by the patient’s estimated plasma volume determined from a nomogram (24). In order to determine if the blood pool concentration derived from PET imaging was similar to that from counting of serum samples, a VOI was drawn over the left atrial blood pool. The SUVmax in the blood was converted to %ID/L of serum using the patient’s hematocrit.

To compare the pharmacokinetics of ⁶⁴Cu-trastuzumab to that of “cold” trastuzumab, serum samples obtained at 30 min and ~24 h were analyzed by ELISA (Genentech, South San Francisco, CA) to determine the concentration of trastuzumab in serum (n=19). One sample was eliminated as erroneous based on an inordinate increase in antibody concentration that biologically was not possible. Since the ⁶⁴Cu-trastuzumab was administered at low levels relative to the therapeutic doses, the ELISA measurements overwhelmingly reflected cold trastuzumab. Measurement of these samples in a gamma well counter provided a determination of the clearance of the radiolabeled trastuzumab. In order to compare the decrease in antibody concentration based on ELISA or radioactivity, the concentration in plasma at 24 h was divided by that at 0.5 h, and the ratios from the ELISA and radioactivity measurements were compared.

Statistics

We performed descriptive statistical analysis as well as a paired t-test to evaluate the differences between groups.

RESULTS

Side effects

All patients tolerated the injections well with no observed side effects during injection or within the 24-hour imaging period.

Pharmacokinetics

Pharmacokinetics of ⁶⁴Cu-trastuzumab based on gamma counting of samples is shown in Supplementary Table 1. The Vc was similar to estimated plasma volume. At the end of infusion, very little tracer had left the plasma volume as determined by multiplying the patients’ estimated plasma volume by the concentration in plasma at the initial sampling time point, as shown in the “amount in PV” in Supplementary Table 1.

Eight patients underwent a repeat ⁶⁴Cu-trastuzumab study. The interval between the first and second study was ranged from 21 to 63 days. The patients had been on a trastuzumab ± chemo regimen for a median of 12 months prior to entering into the PET imaging study (Table 1). The pharmacokinetic parameters between the first and second study were not statistically different, including T_1/2, Co, Vc, and AUC (Table 2). For example, the %ID/L plasma determined by gamma counting at the end of infusion was 42.4 ± 7.8 %ID/L vs. 44.7 ± 12.6 %ID/L for study 1 vs. study 2, respectively (n=8; p=0.45), which was similar to the amount estimated from the fit.

Table 2.

Pharmacokinetic parameters (Study 1 vs. Study 2) *

	Study 1		Study 2
	Mean	SD	Mean	SD	P value
Co (%ID/L) measured	42.4	7.8	44.7	12.6	0.446
Co (%ID/L) fit data	41.4	7.5	44.5	11.4	0.310
Cl (mL/h)	29.5	11.4	38.7	40.9	0.580
Vc (L)	2.473	0.357	2.386	0.634	0.606
Half-life (h)	64.2	22.7	70.7	50.0	0.608
Total area under the curve to last sample (%ID*h /L)	800.1	157.8	715.9	277.9	0.551

^*Co = concentration at time 0

Cl = clearance

Vc = volume of central compartment

%ID*h/L = percent injected dose –h per liter

The clearance kinetics for ⁶⁴Cu-trastuzumab relative to “cold” trastuzumab at 24 vs. 0.5 hour is shown in Supplementary Table 2. The drop in antibody concentration based on gamma counter data was slightly faster than that based on ELISA, with a mean drop of 20% versus 8%, respectively (p=0.001). The retention of antibody based on gamma counting was a mean of 87.7 ± 7% of that based on ELISA (n=18).

Biodistribution

Typical images showed prominent blood pool accumulation at the end of infusion with an estimated 107% of injected dose in plasma volume (Supplementary Table 1 and Figure 1, left upper panel). In the first ~24 hours there was a mean of 21.7 ± 7% drop in plasma activity. The concentration in plasma, based on gamma counting of samples and PET imaging VOI, was not significantly different (Supplementary Table 3). The concentration of ⁶⁴Cu-trastuzumab in plasma measured at early times (2.0 hours ± 0.5 vs. 1.8 hours ± 0.4) were 39.5 ± 7.4 %ID/L vs. 40.3 ± 7.6 %ID/L, respectively, for that measured in the gamma counter (gold standard) vs. PET VOI (p=0.83). The ~24 hour time showed close agreement between the direct plasma gamma count concentration vs. PET estimated concentration (32.3 ± 4.9 %ID/L vs. 35.3 ± 7.7 %ID/L, respectively; p=0.325).

Figure 1.

Patient 8 had localization of both FDG (lower panel) and ⁶⁴Cu trastuzumab (upper panel) in right breast primary. Prominent bloodpool retention was observed.

Visually, blood pool was dominant in the early images and, although slightly decreased at ~24 hours, remained dominant. Liver accumulation was mild and unlikely to decrease sensitivity for detecting liver lesions. Renal and bone accumulation was minor. Whole-body scans performed at ~1.9 hours after injection for studies 1 and 2 showed similar biodistribution in organs (p>0.5), with the exception of small but significantly different SUVlbm in the lung (p=0.02). The biodistribution in study 1 vs. study 2, ~24 hours after injection, showed no statistically significant differences in organ concentration (p>0.05) (Figure 2).

Figure 2.

Plots of mean concentrations of ⁶⁴Cu-DOTA-trastuzumab, expressed as SUV1bm, for all subjects who underwent two series of PET scans (n=8). Upper panel: organ data on day of injection (mean 1.2+/−0.1 h for study 1 vs. 1.2 +/− 0.2 h post-injection for study 2). Bottom panel: organ data for scans performed one day post-injection (mean of 21.6 +/− 3.3 h for study 1 vs. 21.2 +/− 3.3 h post-injection for study 2).

Sites of disease were detected in only 2 patients (patients 1 and 8). Patient 1 had a lesion detected in the liver (Supplementary Figure 1), which progressed at 2.8 months post-initial imaging. Patient 8 had a lesion detected in the breast (Figure 1) and this patient has not progressed with 74-month follow-up on trastuzumab. No sites of disease were detected in the other 9 patients. These 9 patients eventually exhibited progression in various sites (Table 3); in 4 patients, progression occurred in the brain, which was not in the FOV of the ⁶⁴Cu-trastuzumab. In the other 5 patients with negative imaging but progression outside the brain, this occurred at 12.2, 17.5, 8.5, 5.6, and 4 months. The median time on trastuzumab therapy was 38 months, with a range of 2 to 85 months.

Table 3.

Sites of disease, imaging results, trastuzumab therapy history, and follow-up

Patient	Sites of known disease*	Visualized sites 64Cu-trastuzumab	Sites of disease progression	Time to progression from initial 64Cu-trastuzumab injection (months)	Last follow-up since initial antibody (months)	Last trastuzumab treatment since initial antibody (months)
1	Liver+, bone+	Liver	Liver, bone lung	2.8, 7.2, 64.5	90.5	85
2	Lung, bone, brain	Negative	Brain, lung	12.2, 50.6	59.5	58
3	Liver, bone, mediastinum	Negative	Brain	51.6	85.3	85
4	Bone+	Negative	Bone	17.5	42.2	33
5	Bone+, liver−	Negative	Brain	8.2	62.7	38
6	Breast+, bone−	Negative	Breast 5.6m, adrenal 28.2m, liver 33.2m, brain 31m, skin 16.6m	5.6, 28.2, 33.2, 31, 16.6	36.1	35
7	Bone, brain, lung	Negative	Brain	1.4	1.6	2
8	Breast+	Breast	N/A	no progression	74	73
9	Bone+	Negative	Bone, liver, brain, abdominal node, and peritoneal carcinomatosis	4, 17, 23.6, 28.9, 28.9	35.1	34
10	Bone-na	Negative	NA	no progression	70	70
11	Bone-na	Negative	Brain	5.5	32.3	6

^*Note: “+” or “−” refers to patients who had FDG in addition to conventional imaging.

DISCUSSION

HER2 is over-expressed in breast cancer and a variety of other tumors. In spite of these positive results, trastuzumab has activity in only a minority of breast cancer patients with HER2 over-expression, with response rates for trastuzumab monotherapy in the range of 12–26% (25–27). Moreover, clinical resistance to trastuzumab develops in the majority of patients or may present de novo. This clinical observation has motivated the search for other anti-HER2 strategies, including other antibodies, antibody-conjugates, small molecule tyrosine kinase inhibitors of HER2, and agents such as HSP90 inhibitors (2). One of the major obstacles in the clinical development and optimization of these anti-HER2 therapies is the inability to assess the effects of these drugs on their intended targets in vivo. Preclinical studies performed in BT474 breast tumor xenografts by our group showed that ⁶⁸Ga-DOTA-F(ab’)₂-trastuzumab is a better predictor of tumor response than FDG PET (28). Other preclinical studies have shown that radiolabeled trastuzumab may be used to measure on-target effects (29, 30). Furthermore, there has been some indication in the clinic that the ability to target HER2 may help with patient management by predicting patients who are likely to respond to therapy or develop toxicity from therapy (9, 16, 31).

In this study, we successfully chelated trastuzumab with DOTA and labeled it with ⁶⁴Cu, maintaining good immunoreactivity. Our study showed a similar volume of distribution of the central compartment (Vc) of 2.5L compared to 2.95L for repeated doses of cold trastuzumab (32). The clearance estimates were slightly faster in this study -median 25 mL/h compared to 9.4 mL/h in published studies (32). Discrepancies in our measured terminal half-life estimates were a consequence of the short half-life of ⁶⁴Cu, which limited our ability to perform PET scans much later than 24 hours and hence make an accurate determination. The half-life of trastuzumab is known to be dose-dependent and in prior studies, a single exponential fit after administration of 10 mg shows a T_1/2 of 1.1 days (32, 33). When larger administered doses were fit to a single exponential model, the average half-life was 5.8 days.

Measurement of concentration in plasma showed that most of the activity in the initial time was in the plasma volume (Supplementary Table 2) with the expected low clearance. Few studies have compared the clearance of the radiolabeled conjugated antibody with that of the native non-conjugated (34). Our study showed small difference in pharmacokinetics between the ⁶⁴Cu trastuzumab and the unlabeled trastuzumab antibody, probably due to changes from DOTA conjugation or ⁶⁴Cu labeling.

As expected, the PET data from ⁶⁴Cu imaging was quantitative. Data obtained using VOI over the left atrium resulted in estimates of circulating activity that were not significantly different from those based on the ‘gold standard,’ i.e., gamma counting of serum.

The reproducibility of imaging, assessed in 8 patients while on stable therapeutic trastuzumab treatment regimens, was good (Figure 2). In addition, no significant differences were present in pharmacokinetics parameters between study 1 and study 2 based on gamma counting (Table 2). We studied patients receiving chronic trastuzumab in order to evaluate whether the HER2 receptor was completely saturated. Furthermore, since these patients were on chronic fixed doses of trastuzumab, a repeat study would allow us to assess reproducibility.

Lesions were detected in only 2 of 11 patients. The lack of tumor detection in the majority of our patients was not surprising since they were on chronic high-dose treatment of trastuzumab, which was expected to saturate or compete with the small dose of ⁶⁴Cu-trastuzumab. We limited imaging to 24 hours due to the short half-life of ⁶⁴Cu. This 24-hour time point has been shown to be somewhat short for imaging based on ⁸⁹Zr trastuzumab imaging, for which 5 days post-injection was optimal (35). In our own experience, optimal imaging with ⁸⁹Zr trastuzumab in patients with gastric cancer was also at late time points, although we often visualized tumors at 24 hours (personal observation, JAC). Nonetheless, two prior studies utilizing ⁶⁴Cu-trastuzumab documented the feasibility of imaging (19, 20). Mortimer et al found that ⁶⁴Cu-trastuzumab targets metastatic breast cancer in patients not receiving cold trastuzumab therapy. Furthermore, they showed that the majority of the metastatic lesions were detected by 24 hours compared to 48 hours (77% vs. 89%, respectively). The lower sensitivity on day 1 was mainly the result of lesions in the lymph node that were masked by adjacent blood pool and liver.

In contrast to our study, Tamura et al imaged 6 patients with HER2-positive disease; tumors were detected in all 6. Of the 6 patients, 2 were trastuzumab-naïve and the remaining 4 had previous trastuzumab therapy (undisclosed number) with an interval of 1, 1, 8, and 20 days following cold trastuzumab. Although both studies had a small number of patients, the incidence of positive imaging in their patients was higher in their report than ours. This may be due to technical difference in the length of prior cold trastuzumab therapy. In 2 of those treated patients, the site identified was in the brain and it was not certain whether it was non-specific targeting due to blood-brain barrier breakdown or specific targeting. A third patient had visualization of the primary. In that study, a “modest” increase in tumor SUV occurred between 24 and 48 hours. There were significant differences in the design of our study compared to theirs. In our study, patients received a dose of 5 mg in contrast to a mean dose of 86 micrograms in their study. In addition, 2 of their 6 patients had not received prior trastuzumab and rather than administering the ⁶⁴Cu- trastuzumab shortly after the cold dose of trastuzumab, they had an interval of 1 day (2 patients), 8 days (1 patient), and 20 days (1 patient).

The infusion of ⁶⁴Cu-DOTA-trastuzumab was well tolerated. No toxicity or adverse events were associated with its administration, which was expected since this is a very small mass amount of antibody compared to that in their treatment. Similarly, the other two studies using a similar reagent did not report adverse events (19, 20).

While recent studies with ⁸⁹Zr trastuzumab have shown promising results (35), the long half-life imposes limitations such as less favorable dosimetry, limited availability of radionuclide, and patient convenience issues, since patients are required to return ~5 days after tracer administration. Searching for faster targeting radiopharmaceuticals would be useful from a patient convenience point of view, allowing for faster diagnosis and more readily facilitating repeat test/re-test studies as well as assessments following interventions. In conclusion, our study results taken together suggest that this reagent may be used to evaluate HER2 status and pharmacodynamic effects.