Tumor volumes as a predictor of response to the anti-EGFR antibody drug conjugate depatuxizumab mafadotin

Abstract Background The adverse impact of increasing brain tumor size on the efficacy of antibody-drug conjugates (ADCs) was investigated preclinically then validated with clinical data. Methods—Preclinical study The impact of tumor size on ADC tumor delivery and treatment response was evaluated in an EGFR-amplified patient-derived glioblastoma (GBM) model following treatment with Depatuxizumab mafadotin (Depatux-M). Biodistribution and imaging studies correlated drug distribution with starting treatment volume and anti-tumor activity. Methods—Clinical study M12-356 was a Phase I study of Depatux-M in patients with GBM. Blinded volumetric analysis of baseline tumor volumes of M12-356 patients was undertaken by two reviewers and results correlated with response and survival. Results Preclinically, imaging and biodistribution studies showed specific and significantly higher tumor uptake of zirconium-89 labeled Depatux-M (89Zr-Depatux-M) in mice with smaller tumor volume (~98 mm3) versus those with larger volumes (~365 mm3); concordantly, mice with tumor volumes ≤100 mm3 at treatment commencement had significantly better growth inhibition by Depatux-M (93% vs 27%, P < .001) and significantly longer overall survival (P < .0001) compared to tumors ≥400 mm3. Clinically, patients with tumor volumes <25 cm3 had significantly higher response rates (17% vs. 0%, P = .009) and longer overall survival (0.5 vs 0.89 years, P = .001) than tumors above 25 cm3. Conclusion Both preclinical and clinical data showed intra-tumoral concentration and efficacy of Depatux-m inversely correlated with tumor size. This finding merit further investigation with pretreatment tumor volume as a predictor for response to ADCs, in both gliomas and other solid tumors.


Tumor volumes as a predictor of response to the anti-EGFR antibody drug conjugate depatuxizumab mafadotin
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
There is increasing interest in the use of antibodies and antibody-based constructs in glioblastoma (GBM) trials. 1 As antibodies are large macromolecules, with a molecular weight of approximately 150 kDaltons, 2 barriers exist to penetration of brain tumors, even where there are breaches of the blood-brain barrier as seen in enhancing tumors of high-grade gliomas. 3 Concomitantly, other factors such as drug polarity and the presence of complex transport mechanisms may further impede drug penetration. Increased interstitial pressure observed in large tumors is likely one barrier. As GBMs increase in size, larger interstitial pressures develop due to high tumor cell density, increased vascular permeability, and impaired lymphatics. 4 This results in heterogeneous concentrations of drugs in different regions of the tumor. 5 Furthermore, other factors in larger tumors which adversely affect drug uptake including dysfunctional vascular networks and increasing areas of hypoperfusion. 6 We investigated the impact of tumor volume on treatment outcomes using the antibody-drug conjugate (ADC), Depatuxizumab mafodotin (Depatux-M previously ABT-414, AbbVie), both preclinically and clinically. Depatux-M comprises a tumor-specific epidermal growth factor receptor (EGFR) targeting antibody (Depatux, formerly mAb806) linked to the cytotoxin monomethyl auristatin F (MMAF). Importantly, the unconjugated antibody ABT-806 showed no conventional toxicities associated with inhibitors of EGFR signaling, like rash and diarrhea, and imaging of biodistribution of 111 In-ABT-806 showed no normal tissue uptake, highlighting the tumor-specific characteristics of the targeting antibody. [7][8][9] Preclinical data showed in vivo activity of Depatux-M in tumor models with overexpression of wild-type EGFR, EGFR amplification, or EGFRvIII mutation. 10 In the three-arm phase I study (M12-356 study, NCT01800695), Depatux-M given concurrently with radiation therapy and temozolomide in patients with newly diagnosed GBM had an acceptable toxicity profile. 11 In the randomized phase II study, INTELLANCE 2/EORTC 1410, 12 patients with first recurrence GBM were randomised to Depatux-M with or without temozolomide, or lomustine only, or temozolomide only depending on the time of relapse. In this study, the combination of Depatux-M with temozolomide had a 1-year OS rate of 40% versus 28% with lomustine or temozolomide only (HR 0.68, P = .024). No OS difference was observed between Depatux-M monotherapy and temozolomide or lomustine (median OS 7.9 months).
To investigate the impact of tumor size on Depatux-M therapeutic activity, we undertook biodistribution studies using zirconium-89 labeled Depatux-M ( 89 Zr-Depatux-M) to quantitate drug concentration in large and small volume GBMs. We then correlated tumor size with growth inhibition in vivo following Depatux-M treatment. To validate our preclinical findings, we undertook a volumetric analysis of baseline tumor volumes in M12-356 patients and correlated results with patient response as the reduction in drug uptake in larger tumors could be reasonable expected to impact tumor response to Depatux-M. This differential response rate should also be reflected in patient survival, so we undertook to examine the relationship between tumor size and survival.

Preclinical Study
GBM1 tumors were established subcutaneously in NSG mice ( Figure 1). Mice were divided into two groups of 20 each with either small (98 mm 3 ± 20.9 mm 3 ) or large (365 mm 3 ± 0.9 mm 3 ) tumors. In each group, 8 mice received treatment with either Depatux-M or an isotype IgG control ADC, and 2 mice were imaged with either 89 Zr-Depatux-M or 89 Zr-ADC-control (see Figure 1 and Supplementary Material). Mice were injected with 89 Zr-Df-Depatux-M or the 89 Zr-Df-control ADC respectively on the same day as mice in the therapy group commenced treatment with Depatux-M ( Figure 1). Mice from the small size group received between 39-56.2 µg, 41-62.4 µCi 89 Zr-Dfcontrol ADC (n = 2), or 89 Zr-Df-Depatux-M (n = 2) in 100 µl, via tail vein injection. Mice from the large size group received between 38.7-55.3 µg, 30.6-53.2 µCi 89 Zr-Df-ADCcontrol (n = 2) or 89 Zr-Df-Depatux-M (n = 2) via tail vein injection. 89 Zr uptake in normal tissues and tumors was then assessed using PET imaging and MRI performed on a NanoScan PET/MRI (Mediso, Hungary) at the ACRF Centre for Translational Cancer Therapeutics and Imaging (Melbourne, Australia). For biodistribution, all mice were humanely sacrificed by isoflurane over-inhalation after the 168-hour imaging time point and tissues collected for assessment. All animal studies were approved by the Austin Health Animal Ethics Committee and were conducted in

Importance of the Study
Little significant improvement in survival of GBM patients has been made in more than a decade. Many recent trials have investigated antibody and antibody-drug conjugates (ADCs) with disappointing results, often despite success in other tumor types. We show that tumor size directly impacts the deliverable drug concentration of ADCs in brain tumors, and tumor size has a significant impact on therapeutic response. As tumor size can be modulated through surgical or drug approaches, investigating this phenomenon further has implications for the development and use of all large therapeutic molecules in brain tumor patients. In addition, these findings extend beyond GBM as they may be applicable to ADCs and in other tumor types.

Neuro-Oncology Advances
compliance with the Australian Code (8th Edition 2013) for the care and use of animals for scientific purposes.

Clinical Study
To confirm our preclinical data, three reviewers (EL and AS as primary with adjudication as needed by AMS) undertook an independent volumetric analysis of baseline tumor volumes in brain MRI scans of patients treated with Depatux-M on the M12-356 study (Supplementary Figure 2). Manual segmentation of the MRI scans was performed using MIM Maestro TM (MIM Software Inc, Cleveland, OH) under the direction of experienced neuro-oncological radiologists (EL, YP) using 5mm slice thickness of post-gadolinium T1 weighted images (T1wGd) and FLAIR sequences. Precontrast T1 sequences reviewed but not segmented. All nonartefactual FLAIR abnormalities, including suspected edema, were segmented on the FLAIR sequence, with only enhancing disease segmented on T1wGd. The surgical cavity, cysts, and necrosis were not included as per RANO criteria. 13 In cases with multiple disease foci, all regions of interest were segmented. For each case, volumetric data on T1wGd and FLAIR images were documented. Responses were assessed per the RANO criteria in the original M12-356 study, which in brief, incorporate at least 50% reduction in cross-sectional area of contrast enhancement as part of definition of response and at least 25% increase for progressive disease. All clinical data were obtained as part of a clinical trial (NCT01800695) approved by a Human Ethics Committee.

Statistical Analysis
Data from in vivo experiments were analyzed using oneway ANOVA with posthoc Bonferroni correction for TGImax and the Mantel-Cox log-rank test for survival. Analyses were performed using Prism® Version 8.0 (GraphPad, CA). All P-values are two-sided and values ≤.05 were considered significant.
Response rates were compared using chi-square and Fisher's exact tests. Inter-rater correlation and agreement between measurements were assessed using Intraclass correlation coefficient (ICC). 14 ICC estimates were calculated using SPSS version 25 (SPSS Inc. Chicago, IL), two-way mixed-effects, single rater, consistency model. Overall survival was measured from the date of registration to the date of death from any cause. Kaplan-Meier estimates of OS from commencement of therapy were calculated separately for each group and compared using a log-rank test, where P-values ≤ .05 were considered statistically significant.

Therapy Study with 89 Zr-labeled Depatux-M
Half the mice commenced treatment on day 36 when the average tumor volume was <100 mm 3 (98 ± 20.9 mm 3 ) (Treatment at Small Size group). Depatux-M caused  substantial and significant tumour growth inhibition (TGI) compared to control ADC (36.72 ± 6.14 mm 3 vs. 511.02 ± 71.6 mm 3 ; P < .001) with a TGImax of 93%. This result was consistent with multiple previous experiments using these drugs in this model, and other GBM models. 10 The remaining mice were treated on day 71 when their tumor size was 365 mm 3 (Treatment at Large Size group). In this group, Depatux-M also caused significant growth inhibition compared to control ADC but of a lesser magnitude than in smaller tumors (434.95 ± 62.44 mm 3 vs. 625.21 ± 67.28 mm 3 ; P < .01); TGImax was only 27%, which was significantly less than 93% observed in the Treatment at Small Size group (P < .001).  Figure 2A and 2C). By contrast, tumor uptake of 89 Zr-Depatux-M was significantly higher in the Smaller versus the Larger size group (20.93 ± 7.11 ID/g versus 10.67 ± 2.34 %ID/g respectively, P = .0047). In all groups, uptake in liver and bone was consistent with previous literature of 89 Zr-Df metabolism in mice. 15 Whole-body PET/MR images confirmed the biodistribution results, showing a higher uptake of 89 Zr-Depatux-M in the tumors of the Smaller versus Larger size group ( Figure 3A and 3B respectively). Similarly, to the biodistribution data, the corresponding bioimaging study whole-body PET/MR images showed higher uptake of 89 Zr-Depatux-M in Smaller versus Larger size group (13.22 ± 5.27%ID/g versus 6.54 ± 1.11 %ID/g). By contrast, 89 Zr-Control ADC uptake in both the Small Size and Large Size group were low and not significantly different (5.61% ± 0.83 %ID/g versus 4.71 ± 1.58 %ID/g respectively).

Patient characteristics
The M12-356 study (NCT01800695) was an open-label, phase I, 3-arm dose-escalation and expansion study: Arm A evaluated Depatux-M concurrently with radiation and temozolomide (TMZ) in newly diagnosed GBM, Arm B evaluated Depatux-M with TMZ in newly diagnosed GBM as adjuvant therapy as well as in recurrent GBM and Arm C evaluated Depatux-M as monotherapy in recurrent GBM (Supplementary Figure 2). 16

Neuro-Oncology Advances
We obtained baseline MRI scans for all M12-356 patients and undertook a volumetric analysis of tumor volumes at study entry. For Arm A patients, these were the post-operative and preradiotherapy MRI scans. A total of 202 patients were included in the current analysis. Baseline demographic data are detailed in Table 1. The median age was 57 years, majority were male (59%) and majority (87%) had a Karnofsky performance scale ≥ 80. Tumors were EGFR amplified in 71% of patients and EGFRvIII mutant in 52%. Only 13% of those tested were MGMT promoter methylated.

Efficacy
We showed that there was excellent inter-rater correlation for volumetric analysis using T1wGd and FLAIR images (kappa = 0.89, P < .0001 and kappa = 0.97, P < .0001 respectively). Importantly, we found that patients with EGFR amplified recurrent GBMs (n = 110) treated with Depatux-M, either alone (Arm C) or in combination with TMZ (Arm B), had significantly more responders in patients with tumor volumes < 25 cm 3 compared to those ≥25 cm 3 at study entry (response rate of 17% vs 0%, P = .009 two-sided). In fact, no patient with a ≥25 cm 3 had a RANO response. When analyzing by arm, responses were also more frequent in smaller than larger tumors in Arm B (50 patients who received Depatux-M with temozolomide, response rate 24% vs 0%, P = .089 two-sided) and Arm C (60 patients who received Depatux-M alone, response rate 10% vs 0%, P = .287) but did not reach statistical significance in this underpowered posthoc exploratory subset analysis. Of note, most patients in this study had recently progressed (within 6 months) on TMZ and the likelihood of response to TMZ alone in this patient population would be low. 17 Patients in Arm A were not considered evaluable for response due to recent completion of radiotherapy.
We also examined the impact of tumor volume on survival. Patients with newly diagnosed GBM (n = 41) enrolled on arms A and B whose tumor volumes were below 25 cm 3 (post-operatively but before chemoradiotherapy), had a significantly longer median OS than those above 25 cm 3 (2.0 vs 0.8 years; P = .006) respectively) ( Figure  4A). In this cohort, analysis in the subset of patients with EGFR amplified tumors (n = 16) trended toward improved survival for patients with tumors volumes <25 cm 3 than > 25 cm 3 (1.8 vs 0.8 years, P = .28) ( Figure  4B). Patients in Arms B and C with EGFR amplified recurrent GBMs (n = 110) with tumors < 25 cm 3 had a longer median survival than patients with tumors ≥25 cm 3 (0.81 vs 0.52 years, P = .001) ( Figure 5A). This association of tumor volume and survival was also seen in patients on Arm C (n = 60), with tumor volumes < 25 cm 3 significantly associated with longer OS (0.89 vs 0.50 years, P = .001) ( Figure 5B).

Discussion
To our knowledge, this is the first study to quantify the impact of tumor volume on the uptake, retention, and therapeutic efficacy of ADCs in brain tumors. Data to date would suggest that large tumors in other sites are associated with increased interstitial pressure, abnormal vasculature and lymphatics, increased rates of necrosis, and increase heterogeneity. [4][5][6]19 For example interstitial pressures in center of large tumors may be four-fold higher than for small tumors. 19 Our results strongly support our hypothesis that larger tumor size, with associated adverse features expected to reduce drug penetrance and uptake, resulted in significant reductions in the uptake of Depatux-M in preclinical models, with corresponding reduction in tumor growth inhibition. Clearly, this raises the possibility of the potential confounding influence of tumor volume on assessment of efficacy of Depatux-M. However, these findings have implications for the optimal integration of all ADCs or other large molecules in the management of GBM, strongly supporting strategies that would reduce tumor size and/or interstitial pressure to increase their efficacy. Furthermore, tumor size is rarely controlled for in many GBM trials and may contribute to underestimation of drug efficacy. Several strategies to mitigate the impact of larger tumors are potentially available. Bevacizumab results in vascular normalization, reduces vascular permeability, and has been shown to impact intra-tumoral drug distribution and potentially efficacy of ADCs when used in combination. 20 For example, platinum-resistant epithelial ovarian cancer xenograft models treated with the ADC Mirvetuximab soravtansine (IMGN853), a FRα-binding antibody linked to a tubulin-disrupting maytansinoid (DM4), in combination with bevacizumab resulted in significant tumor regression that was superior to either bevacizumab or the ADC alone in ovarian cancer. 21 It was postulated the presence of bevacizumab results in better tumor penetration and exposure to the ADC, resulting in more effective eradication of tumor cells. These findings support investigating anti-angiogenic and ADC combinatorial approaches to further enhance the therapeutic benefit of these agents. The ongoing phase Ib FORWARD II trial (NCT02606305) is evaluating the Mirvetuximab soravtansine in combination with bevacizumab in pts with platinum-resistant ovarian cancer. 22 Another approach would be simple debulking of the tumor prior to ADC treatment. Surgery at recurrence has shown to be associated with a survival advantage and has not shown to significantly affect the quality of life. [23][24][25][26] With evolving surgical techniques and better patient selection, more patients are then able to receive systemic therapy postoperatively. 25 A number of other experimental pharmacological and physical strategies have been investigated to reduce tumoral interstitial pressure including imatinib, paclitaxel, dexamethasone, angiotensin II, TGFβ inhibitors as well as hyperthermia, radiotherapy, photodynamic therapy, and focused ultrasound therapy. 27,28 Clearly, these findings require further validation, ideally in orthotopic models and including other drug treatment, followed by a prospective clinical trial if appropriate. We are currently undertaking additional work to investigate how tumor volume relates to traditional prognostic and predictive biomarkers in this nonrandomized phase 1 study. We are also currently seeking to confirm our findings with data from Depatux-M in the INTELLANCE 1 and 2 studies of Depatux-M in newly diagnosed and recurrent GBM, respectively, where the randomized designs would allow us to differentiate between the predictive impact of tumor volume on survival based on differential tumor

Neuro-Oncology Advances
efficacy compared to the well-known prognostic effects of tumor volume. [29][30][31][32] These analyses will provide us with definitive data as to whether tumor volume impacts drug delivery and patient outcomes. If our initial findings are confirmed, it will require a significant change in how future research and clinical trials are designed.

Conclusion
Increased tumor volumes result in significant reduction in ADC penetration in GBM preclinically and response in patients. The impact of this as a modifiable factor, within the broader prognostic impact of increased tumor volume, warrants further investigation with prospective and/or randomized trials.

Supplementary Material
Supplementary material is available at Neuro-Oncology Advances online.