Phase I/II Study of the Mesothelin-targeted Immunotoxin LMB-100 with Nab-Paclitaxel for Patients with Advanced Pancreatic Adenocarcinoma

Pancreatic cancer is an aggressive malignancy with a 5-year overall survival of only 9% despite recent advances in combination chemotherapy treatments (1–3). In fact, pancreatic cancer is now the third most common cause of cancer-related death in the United States (4). Pancreatic ductal adenocarcinoma (PDAC) accounts for approximately 90% of this disease burden. Single agent immune checkpoint inhibitor treatments, targeted mAbs, and therapies directed against receptor tyrosine kinases have been clinically ineffective for this disease, except in rare cases (5). New research has begun to show that patients with PDAC harboring BRCA mutations (∼5%–7%) may respond to PARP inhibition (6–9), but personalized therapies directed against tumor-associated mutations has been disappointing for most other patients with PDAC as the most commonly mutated genes in PDAC (KRAS, TP53, DPC4, and CDKN2A) remain elusive drug targets (10). Novel therapies are needed to treat PDAC.

LMB-100 (previously called RG7787 and Ro6927005) is a second-generation immunotoxin (iTox) designed to kill cancer cells that express the cell surface glycoprotein mesothelin (MSLN). MSLN is a differentiation antigen for the mesothelial cells of the pleura, pericardium, and peritoneum, but is also expressed by >90% of PDAC (11). LMB-100 consists of a humanized Fab that binds to MSLN with high affinity fused to a modified Pseudomonas exotoxin A (PE) payload (12). LMB-100 is directed to MSLN-expressing cells by the binding domain, and then internalized through endocytosis, resulting ultimately in cytoplasmic release of PE. PE is an enzyme which kills cells by irreversibly modifying elongation factor-2 to halt protein synthesis, a unique mechanism of action. Preclinical studies demonstrated LMB-100 antitumor activity in mouse models of PDAC and other MSLN-expressing solid tumors (13, 14). Phase I testing of LMB-100 identified a single agent MTD of 140 μg/kg. Dose-limiting toxicities (DLT) included capillary leak syndrome (CLS) and reversible elevations of creatinine.

Combination studies in the laboratory have identified synergistic antitumor effect when LMB-100 is combined with other anticancer drugs including dactinomycin (15), panobinostat (16), and taxanes (13, 17). Complete responses were observed in a pancreatic tumor model when mice were cotreated with LMB-100 and nanoalbumin bound (nab)-paclitaxel (17). Given these results, we conducted a clinical trial examining the safety and antitumor response of this combination in patients with PDAC.

Eligible patients were ≥18 years old and had a histologically confirmed diagnosis of PDAC. Advanced or recurrent disease previously treated with at least one line of standard chemotherapy was required. Prior nab-paclitaxel was permitted if >4 months since last administration. Other requirements included: measurable disease per RECIST version 1.1, Eastern Cooperative Oncology Group performance status (ECOG PS) 0 or 1, adequate organ function including baseline documentation of left ventricular ejection fraction ≥ 50% by echocardiogram, and ambulatory oxygen saturation > 88%. See full protocol in Supplementary Materials and Methods for full inclusion and exclusion criteria. The study was conducted in accordance with FDA regulations and Good Clinical Practice guidelines. The study protocol was approved by the NCI Institutional Review Board and written informed consent was obtained from all patients participating.

This open-label, phase I study of intravenous LMB-100 was conducted at NCI Center for Cancer Research (Bethesda, MD; NCT02810418). Results for all study arms where patients received short infusion LMB-100 (30 minute infusion) with nab-paclitaxel (Arms A1 and A2) are reported here. This schedule of LMB-100 was given in prior single agent phase I testing (NCT02317419 and NCT02798536). Data pertaining to patients that received long infusion of LMB-100 (infusion > 24 hours), an alternative administration schedule, as a single agent (Arm B1) or with nab-paclitaxel (Arm B2) will be reported separately. Arm B1 was enrolled concurrently with the A Arms (Supplementary Fig. S1). Patients ineligible for Arm A or unwilling to receive nab-paclitaxel were enrolled on Arm B1. Following completion of Arm B1, enrollment on Arm A was halted pending full accrual of Arm B2. The primary endpoint for Arm A1, the phase I dose escalation, was determination of a MTD for LMB-100 when given with nab-paclitaxel. For the phase II expansion (Arm A2), objective response rate (ORR) by RECIST criteria was the primary endpoint. Secondary endpoints included assessment of adverse events (AEs), determination of pharmacokinetics, incidence of antidrug antibodies (ADA) against LMB-100, and change in tumor marker CA 19-9. The phase II expansion was conducted using Simon minimax two-stage phase II trial design to rule out an unacceptably low ORR of 5% (P₀ = 0.05) in favor of an improved ORR of 25% (P₁ = 0.25) with α = 0.10 and β = 0.10. Planned enrollment would include 20 evaluable participants (including those treated at MTD on the phase I) through the second stage only if at least 1 of the first 13 participants experienced a response. Participants were considered evaluable for phase II if they had received at least one dose of LMB-100, had completed posttreatment imaging studies, or were taken off-treatment for clinical progression of disease. The study was voluntarily closed to accrual by the study team during the phase II expansion after enrollment of 14 participants at MTD due to unacceptable toxicity.

A 3+3 design was used for the dose escalation. DLT was defined as most grade ≥ 3 nonhematologic toxicity (except for clinically insignificant electrolyte abnormalities) and specified severe hematologic toxicities (see Supplementary Materials and Methods for full protocol) that were related to LMB-100 and occurred within 21 days of the first LMB-100 administration. Nab-paclitaxel was administered at 125 mg/m² on days 1 and 8, and LMB-100 by 30-minute infusion as per dose escalation on days 1, 3, and 5 of each 21-day treatment cycle. This schedule was chosen to match the 21-day cycle length of prior LMB-100 single-agent studies (NCT02317419 and NCT02798536), to correspond with preclinical data on the efficacy of the regimen in mice (17), and to minimize interruptions in treatment due to myelosuppression from nab-paclitaxel in this heavily pretreated patient population. On day 1 (when both drugs were given), nab-paclitaxel infusion preceded LMB-100 by 30 minutes. This schedule was chosen because preclinical studies performed in mice demonstrated increased toxicity if nab-paclitaxel was infused immediately after LMB-100 or between the LMB-100 doses. All patients were premedicated with acetaminophen, diphenhydramine, and ranitidine prior to LMB-100 infusion and ondansetron was available as needed for nausea. Initially, up to four treatment cycles were permitted, however, a two-cycle limit was introduced by protocol amendment due to frequent incidence of infusion-related reaction beyond cycle 2. Dose reductions of nab-paclitaxel were permitted as per package insert. LMB-100 used for this study was manufactured by Roche then transferred to NCI. Nab-paclitaxel was obtained from commercial sources.

AEs were graded using Common Terminology Criteria for Adverse Events (CTCAE) version 4.0. AE of CLS was not routinely recorded. Instead, individual AEs associated with CLS were each recorded separately from that designation to better understand the clinical symptoms associated. Baseline CT or MRI was performed within 28 days of treatment initiation with follow-up scans after every two cycles of treatment (∼6 weeks). Tumor response was assessed per RECIST version 1.1. CA 19-9 tumor marker was drawn prior to treatment on the first day of each cycle and during follow-up visit after the last study treatment.

LMB-100 drug concentrations were assessed from plasma drawn predose, end of infusion (EOI), and 1, 2, 3, 4, and 6 hours post-EOI. Free LMB-100 plasma concentrations were measured with a validated ELISA with a lower limit of quantification (LLOQ) of 2.1 ng/mL. Testing was performed by contract with Frederick National Laboratory for Cancer Research operated by Leidos Biomedical Research, Inc. For nab-paclitaxel assessment, blood samples were drawn predose, 15 minutes post start of infusion, 5 minutes prior to EOI, EOI, and 0.5, 1, 2, 4, 8, 12, 24, 48, and 72 hours post-EOI. Briefly, total paclitaxel (bound to nab-paclitaxel particle, freely dissociated, and bound to plasma proteins) was measured by a validated ultra-high-performance HPLC/MS-MS assay, as described previously (18) with one small modification to decrease the LLOQ from 10 ng/mL to 5 ng/mL. Briefly, 5× volume of acetonitrile was added to 100 μL of plasma containing nab-paclitaxel to precipitate proteins and release any bound paclitaxel. The mixture was vortexed and filtered via 0.2-μm Varian Captiva 96-well filtration plate. The supernatant was injected onto a Waters SymmetryShield RP18 column, 2.1 × 50 mm, 3.5 μm before mass spectrometric detection and quantification of total paclitaxel plasma concentrations in the positive ion mode.

Pharmacokinetic parameters, namely maximum plasma concentration (C_max), AUC extrapolated to time infinity (AUC_INF), half-life (T_1/2), clearance (CL), and volume of distribution (V_z), were estimated using noncompartmental methods for both LMB-100 and nab-paclitaxel separately. Phoenix WinNonlin v8.1 was used to perform pharmacokinetic analyses; all plots and statistical tests were performed in GraphPad Prism, v8. The nonparametric Mann–Whitney test was used to compare group means for each parameter based on dose (65 vs 100 μg/kg).

A validated screening ELISA assay was applied to plasma samples drawn prior to drug administration on day 1 of each cycle and at posttreatment follow-up visit. Samples with mean assay signals greater than the cut-off point (OD 0.05) were run in a confirmatory assay where a percent inhibition by spiked LMB-100 (32 μg/mL) was calculated. Inhibition of mean assay signal by >41.8% was used as threshold to define confirmed positives. Testing was performed by contract with Frederick National Laboratory for Cancer Research operated by Leidos Biomedical Research, Inc.

Blood was collected in BD Vacutainer Cell Preparation Tubes with sodium citrate (BD Biosciences). Peripheral blood mononuclear cells were isolated, aliquoted, and viably frozen and stored in liquid nitrogen until use. CEC analyses were performed using a MACSQuant flow cytometer (Miltenyi Biotec) as described previously (19). Briefly, a minimum of 1 × 10⁵ cells were acquired for each analysis. CECs were defined as negative for the hematopoietic marker CD45 (leukocyte common antigen), positive for the endothelial markers CD31 and CD146, and negative for the progenitor marker CD133. CECs were further subgrouped into viable versus apoptotic populations. Viability was defined by the absence of 7-aminoactinomycin D staining, and analysis was restricted to nucleated cells by gating on Hoechst 33342–positive cells. Data were analyzed using FlowJo Software (FlowJo LLC).

Graphs were generated using Microsoft Excel or GraphPad Prism 7.01 and statistical analyses were performed in Prism or using SAS version 9.4 (SAS Institute, Inc.). Spearman correlation was performed to demonstrate the association between change in percentage of apoptotic CEC and percent weight gain from baseline to cycle 1, day 5. A LOESS line was fit to these data to show their relationship. Error bars show SDs unless otherwise indicated. The safety analysis included all patients who received at least one dose of LMB-100.

Twenty patients were enrolled to this study between August 2016 and February 2019 and assigned to receive short infusion LMB-100 (Arms A1 and A2) in combination with nab-paclitaxel. Fourteen were treated on the dose escalation (Arm A1) and six were part of the phase II expansion at MTD (Arm A2). Patients received a median of two cycles (range, 1–3) of study treatment. Demographics, treatment history, and disease burden of these patients are outlined in Table 1. Notably, 35% of patients had received prior nab-paclitaxel, 65% had liver metastases upon enrollment, and 20% had clinically apparent ascites.

Six patients received 100 μg/kg LMB-100 (dose level 1) with nab-paclitaxel. Two of 6 patients developed DLT of myalgia (grade 3) with weight gain >5 kg due to edema from CLS. Myalgia manifested as severe, narcotic-refractory pain that inhibited ambulation but was not associated with serum creatinine kinase elevation. MRI of the thighs in one of these patients showed fascial edema but no other abnormality of the muscle to suggest myopathy or myositis. Other CLS-associated DLTs of fatigue (grade 3) and hypotension (grade 3) were also recorded in these 2 patients.

Because unacceptable toxicity was encountered at the first dose level, the LMB-100 dose was reduced to 65 μg/kg (dose level −1). Eight patients were treated at this dose level. Two patients who did not experience DLT but required a hold or dose reduction of the day 8 nab-paclitaxel for reasons unrelated to LMB-100 treatment were replaced for the DLT analysis. Of 6 evaluable patients at this dose-level, CLS-associated DLTs of grade 3 edema and grade 3 urine output decreased were observed in one participant. The 65 μg/kg dose was therefore advanced to phase II testing.

Figure 1 shows AEs attributed to LMB-100. AEs experienced by those treated at the MTD of 65 μg/kg on both the phase I and phase II are tabulated in composite. The most common AEs were hypoalbuminemia (100%), edema (65%), fatigue (50%), hyponatremia (50%), and myalgia (45%). Grade 4 toxicities of ventricular dysfunction, hypotension, and atrial fibrillation occurred in patient No. 40 who was treated at the MTD on the phase II. Cycle 1 day 5 LMB-100 was held in patient No. 40. There were no other dose holds or dose reductions of LMB-100. Six patients discontinued treatment after cycle 1 due to LMB-100 toxicity.

All AEs that were attributed to nab-paclitaxel are reported in Supplementary Fig. S2. The most common events were decrease in blood counts (anemia, lymphopenia, leukopenia, neutropenia, and thrombocytopenia). Fatigue, nausea and vomiting, and diarrhea were also observed. These are consistent with the known toxicities of nab-paclitaxel. During cycle 1, three patients required hold of day 8 nab-paclitaxel. Four patients required dose reduction of nab-paclitaxel during cycle 2 due to prior neutropenia or worsening peripheral neuropathy. Severe AEs (grade 3–5) attributed to disease are reported in Supplementary Fig. S3. All were associated with non-neutropenic infections.

CLS was the most common AE associated with LMB-100 treatment. All participants experienced at least one symptom related to CLS (Fig. 2A). To better define the scope and timing of symptoms that patients with LMB-100–associated CLS experienced, we documented occurrence of each individual symptom rather than using the blanket CLS CTCAE term. Three participants were excluded from this analysis due to the development of sepsis requiring aggressive fluid resuscitation during cycle 1, a treatment that confounded attributions. Hypoalbuminemia occurred in all participants and began by day 3 (Fig. 2B). Serum albumin continued to drop through day 8 in most patients then began to recover after day 15. Weight gain from fluid edema was observed in most participants (Fig. 2C). Increases in weight began around day 5 and peaked between days 8 and 15. Nine of the 17 participants considered evaluable (53%) developed mild CLS, which was defined as CLS resulting in <5 kg weight gain from edema (Fig. 2A). AEs associated with mild CLS included hypoalbuminemia (grade 1 or 2), hyponatremia (grade 1), and mild edema, which resolved spontaneously, and asymptomatic hypotension. The remaining eight participants (47%) developed significant CLS (grade 2–4, or not classified as CLS per CTCAE criteria) with weight gain of 5 kg or more. Dizziness (4/8), tachycardia (4/8), edema requiring outpatient diuresis with oral furosemide (6/8), and hyponatremia (6/8, including 3 with grade 3) were also observed in this population. Four of these participants developed serious CLS (grade 3 or 4) that required hospitalization for symptom management such as diuresis with intravenous furosemide for refractory edema or fluid resuscitation for hypotension. One patient developed life-threatening CLS (grade 4) due to edema of the ventricular walls, which decreased left ventricular ejection fraction (Fig. 2D). The impairment in cardiac function caused worsening hypotension and cardiogenic shock requiring vasopressor support for approximately 7 days until CLS resolved. No patient-specific factors that could have predicted occurrence of these life-threatening AEs were identified. The trial was closed to accrual of new patients in response to this severe AE.

CLS induced by iTox is presumed to be caused by nonspecific uptake of iTox by endothelial cells resulting in their direct damage. We analyzed CECs from patient blood samples drawn prior to treatment and within 24 hours of the cycle 1 day 5 LMB-100 dose. We found a moderately strong correlation between increase in apoptotic CECs and severity of CLS as measured by percent increase in body weight following treatment (Fig. 2E, r = 0.68, 95% confidence interval, 0.27–0.88). There was no association between % change in viable CECs (Supplementary Fig. S4) or baseline kidney function (Fig. 2F) and severity of CLS.

There were 20 patients with nab-paclitaxel serum concentration data available. Mean nab-paclitaxel plasma concentration followed a biphasic disposition (Fig. 3A), consistent with a prior report (20). Individual parameter estimates, along with a statistical summary based on sex were also determined (Supplementary Table S1). It is difficult to compare the C_max and AUC parameters with previously published literature because a different dose of nab-paclitaxel was administered. However, the T_1/2, CL, and V_z were very similar to results shown in previous pharmacokinetic analyses of nab-paclitaxel in humans (20).

LMB-100 plasma concentrations were available for all 20 participants and declined in a mono-exponential manner following a short 30-minute intravenous infusion (Fig. 3B). On cycle 1 day 1, patients given 100 μg/kg had significantly higher C_max (P = 0.012) and greater AUC (P = 0.006) than those given 65 μg/kg, but without significant differences in dose-normalized C_max and AUC, as well as T_1/2, V_z, and CL (Table 2).

During cycle 1, C_max greater than 100 ng/mL was measured for all 20 participants (Fig. 3C). A wide distribution was observed for this parameter at MTD with values ranging from 140 to 1538 ng/mL. We hypothesized that this variation could be due to the presence of preexisting ADAs. Indeed, we found a modest inverse correlation between baseline ADA value pretreatment and cycle 1 C_max (Fig. 3D). During cycle 2, eight of 13 participants had C_max > 100 ng/mL (62%), but five had very low drug levels (Fig. 3C). All five with low drug levels had a positive ADA screening test at baseline (Fig. 3E). However, three of the eight participants with C_max > 100 ng/mL also tested positive for ADAs before starting treatment. Moreover, two participants with positive ADA testing performed immediately after pretreatment on C2D1 also achieved C_max > 100 ng/mL for cycle 2. While a negative ADA test was predictive of achieving good drug levels, a positive ADA test did not predict poor cycle 2 drug levels. Post cycle 2, ADA values were available for eight participants. Seven of eight (88%) developed high titer ADAs after receiving six doses of LMB-100. This suggests that few patients would achieve good drug levels if more than two cycles of LMB-100 were administered.

CA 19-9 response was evaluable for 17 participants. Maximum decline in CA 19-9 of >50% was observed in seven participants during their two cycles of treatment with LMB-100 and nab-paclitaxel (Fig. 4A). Three of seven with CA 19-9 response had received prior nab-paclitaxel. CA 19-9 normalized in patient No. 15. This patient also experienced a confirmed partial response with 100% decrease of a FDG-avid mass anterior to the liver (Fig. 4B). Patient No. 15 continued to have imaging abnormalities consistent with tumor near the pancreatic head so could not be classified as a complete responder. In addition, patient No. 25 who had received prior nab-paclitaxel and had liver-predominant disease experienced a 17% reduction in tumor burden. These two participants with the best imaging responses both had cycle 2 LMB-100 C_max > 1,000 ng/mL.

IHC to assess tumor MSLN expression was performed on archival tumor samples. These samples were available for 10 of 20 participants, including six of the seven with CA 19-9 response to treatment (Supplementary Table S2). All participants with CA 19-9 response had MSLN expression in 40% or more of their tumor cells (Fig. 4A). The four participants with MSLN IHC data available and worst CA 19-9 responses expressed MSLN in 30% or less of their tumor cells. Data from this limited number of patients suggests that CA 19-9 response to combination treatment with LMB-100 and nab-paclitaxel may correlate with higher tumor MSLN expression.

Our data show that the combination of nab-paclitaxel with LMB-100 may have antitumor activity, but amplifies the toxic side effects of the iTox. The combination caused severe symptoms of CLS even when LMB-100 was administered at less than 50% of the single-agent MTD. In addition, unique manifestation of CLS was observed when LMB-100 was combined with nab-paclitaxel, such as severe myalgia and cardiac toxicity. These severe toxicities outweighed the benefits of treatment with this combination.

Here, we have reported detailed clinical information on the manifestation of iTox-associated CLS. This syndrome begins with albumin loss occurring within 48 hours of iTox infusion, followed by weight gain from edema that peaks by 2 weeks into the cycle. All CLS-related toxicities were fully reversible with the administration of adequate supportive care. No long-term changes in renal or cardiac function were observed. The etiology of iTox-induced CLS is not well understood. Early-generation iTox's targeting Lewis antigens resulted in CLS at least in part due to specific uptake of iTox by Lewis antigen–expressing endothelial cells. However, for iTox, like LMB-100, that target a protein not expressed by endothelial cells another mechanism must be responsible. While nonspecific uptake of LMB-100 by endothelial cells is possible, in vitro testing using cultured endothelial cell lines has shown that exposure to concentrations of iTox that far exceed our observed C_max is insufficient to induce cell killing unless the cells express the iTox target (21). Nevertheless, we found that apoptotic CECs increased in all patients immediately following LMB-100 treatment. Furthermore, a quantitative association between severity of CLS and numbers of apoptotic CECs was observed. Further study will be required to determine whether this is due to off-target uptake of LMB-100 by endothelial cells causing endothelial cell killing or whether a secondary process such as iTox-induced inflammation produces this damage. In addition, we do not know whether coadministration of nab-paclitaxel is necessary to trigger the observed increase in apoptotic CECs.

While the combination of nab-paclitaxel with LMB-100 will not be pursued further, other clinical trials testing LMB-100 are planned or continue (NCT03644550). Several of our observations pertaining to companion diagnostics are relevant to these studies. First, our data have shown that patients with the best CA 19-9 response to LMB-100/nab-paclitaxel tested positive for MSLN expression in ≥40% of tumor cells in archival tumor samples, while nonresponders all had expression less than this. Second, we have found that all patients with a negative ADA test achieved expected circulating LMB-100 levels during second cycle treatment. Third, a positive ADA test did not guarantee poor circulating LMB-100 levels during cycle 2. These data provide an initial indication of which patients are most likely to benefit from LMB-100 if corroborated by additional patient data.

In summary, the LMB-100 and nab-paclitaxel combination did show evidence of clinical activity but was too toxic for patients with pancreatic cancer. Future LMB-100 combination studies with other drugs are planned.

ADAs against LMB-100 limit the amount of drug that can be given. Toxicity of this combination makes it unsuitable for further development.

R. Hassan has received funding for conduct of clinical trials via a cooperative agreement between NCI and Bayer AG, Aduro BioTech, and Morphotek Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Alewine, R. Hassan, I. Pastan

Development of methodology: C. Alewine, M.-J. Lee, A. Thomas, J.B. Trepel, W.D. Figg, R. Hassan, I. Pastan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Alewine, M. Ahmad, M.-J. Lee, A. Yuno, J.D. Kindrick, A. Thomas, J.B. Trepel, W.D. Figg, R. Hassan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Alewine, C.J. Peer, Z.I. Hu, M.-J. Lee, J.D. Kindrick, A. Thomas, S.M. Steinberg, J.B. Trepel, W.D. Figg, R. Hassan

Writing, review, and/or revision of the manuscript: C. Alewine, C.J. Peer, Z.I. Hu, M.-J. Lee, A. Thomas, S.M. Steinberg, J.B. Trepel, W.D. Figg, R. Hassan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Ahmad

Study supervision: C. Alewine, M. Ahmad

The authors would like to thank all the patients who participated in this study and their caregivers for supporting pancreatic cancer research, you are always in our hearts. We would also like to thank Paula Carter in the Blood Processing Core, Yanyu Wang and Jon Inglefield at Frederick National Laboratory (Leidos Biomedical Research, Inc), and research nurses Yvonne Mallory and Lisa Bengtson. This study was sponsored and funded by the Intramural Research Program of the NIH, NCI, Center for Cancer Research Intramural Research Program (project no. ZIA BC 011652).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.