Erlotinib Pretreatment Improves Photodynamic Therapy of Non–Small Cell Lung Carcinoma Xenografts via Multiple Mechanisms

A leading characteristic of aggressive tumors is aberrant expression of the epidermal growth factor receptor (EGFR), a signaling protein involved in promoting cell proliferation, angiogenesis, and survival. Several EGFR-inhibiting drugs are used to treat malignancies such as non–small cell lung carcinoma (NSCLC), head and neck squamous cell carcinoma (HNSCC), and breast cancer (1, 2). These drugs include EGFR-targeting antibodies (i.e., cetuximab) and small-molecule inhibitors (e.g., gefitinib and erlotinib). The efficacy of EGFR inhibition varies based on the molecular characteristics of the tumor. For example, tumors harboring EGFR-activating mutations tend to be more sensitive to EGFR-targeted therapy, whereas tumors that exhibit KRAS mutations respond poorly; and while some tumor types are initially sensitive to EGFR inhibition, others will acquire resistance, leaving patients with limited treatment options (1, 3). Thus, while EGFR has gained favor as a molecular target for treating several types of tumors, inhibition of the receptor as standalone treatment does not consistently yield clinically positive results.

Photodynamic therapy (PDT) uses a combination of light, photosensitizer, and oxygen to inflict cell damage. Combinations of PDT with drugs that target EGFR suggest that multimodality therapy can increase cytotoxicity (4, 5). EGFR targeting of photosensitizer reduces cell proliferation after PDT (6), and anti-EGFR antibodies and small-molecular inhibitors can increase apoptosis and/or inhibit proliferation in tumors or cell cultures that are treated with PDT (7–10). Therapeutic benefit has been shown in combinations of PDT with cetuximab in tumor models of ovarian, nasopharyngeal, and bladder cancer (7–9). Such results have stimulated clinical interest in combining EGFR inhibition with PDT, but the mechanisms that underlie this added therapeutic benefit are poorly understood.

The goals of this study are 2-fold. First, we establish the therapeutic efficacy of adding erlotinib, a small-molecule inhibitor of EGFR, before PDT of NSCLC human tumor xenografts. Erlotinib is an oral drug, so pre-PDT administration is translatable to intraoperative delivery of PDT, where oral drug administration to patients in the intensive care unit postoperatively is not feasible. Second, we have elucidated the mechanisms by which response is improved using this protocol. The effects of erlotinib on PDT-induced tumor cytotoxicity, endothelial cell damage, and vascular function have been determined and studied for their contribution to long-term PDT responsiveness. These findings provide insights that will inform the clinical development of treatment regimens that incorporate targeting of EGFR with PDT.

EGFR wild-type, KRAS-mutant human NSCLC cell lines, H460 and A549, were obtained from the ATCC. H460-mCherry cells and SVEC (immortalized mouse endothelial cells) were provided by Drs. Jay Dorsey (University of Pennsylvania, Philadelphia, PA) and Bin Chen (University of the Sciences, Philadelphia, PA), respectively. Cells were maintained in RPMI-1640 (H460, SVEC) or DMEM (A549) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. All cell lines were authenticated by and confirmed as Mycoplasma free using the Cell Check Plus authentication service from IDEXX Bioresearch (May 2014).

Tumors were propagated by intradermal injection of 1 × 10⁶ cells over the right shoulder of female nude mice (Athymic, Ncr-nu/nu; National Cancer Institute, Frederick, MD). Studies were initiated when tumors reached diameters of 6 to 8 mm. Investigations were approved by the University of Pennsylvania Institutional Animal Care and Use Committee; animal facilities are Association for Assessment and Accreditation of Laboratory Animal Care accredited.

Animals received benzoporphyrin derivative monoacid ring A (BPD, Visudyne; QLT Ophthalmics), at 1 mg/kg (i.v.), 3 hours before light delivery (690 ± 5 nm). Light from a diode laser (B&W Tek, Inc.) was delivered through microlens-tipped fibers (Pioneer) to produce a 1 cm diameter field of uniform intensity. Laser output was measured (LabMaster power meter; Coherent) and adjusted to produce a fluence rate of 75 mW/cm² at the tumor surface. A fluence of 135 J/cm² was delivered, during which mice were anesthetized by inhalation of isoflurane in medical air (VetEquip anesthesia machine).

PDT was combined with erlotinib (Tarceva, Genentech) or bevacizumab (Avastin, Genentech). Erlotinib (50 mg/kg) was administered (oral gavage) at either 3 hours before PDT (single dose) or at 2 days, 1 day, and 3 hours before PDT (triple dose). Dosing was chosen from pilot studies in which 50 mg/kg was efficacious in the absence of toxicities, whereas higher doses (up to 100 mg/kg) resulted in dramatic weight loss in the animals. Bevacizumab (10 mg/kg) was administered via intraperitoneal injection at 2 days, 1 day, and 3 hours before PDT.

After treatment, animals were followed daily to assess long-term outcome. Tumor volume was calculated using the formula volume = diameter × width² × π/6. Animals were euthanized when tumor volume exceeded its pretreatment measurement or at 90 days after PDT in the absence of tumor recurrence (i.e., complete response).

Phosphorescence lifetime-based measurements of tumor oxygenation were performed as described previously (11) using intratumoral injection of the probe Oxyphor G4 (20 mmol/L). Erlotinib was administered on a triple-dose schedule. Tumor pO₂ was measured before erlotinib (50 mg/kg) administration on each day and then at 3 hours after the final dose (corresponding to measurements at 0, 24, 48, and 51 hours relative to the first dose of erlotinib). Phosphorescence lifetime oximetry is based on variations in the probe phosphorescence decay time due to quenching of the probe triplet state by oxygen. The measured quantity (phosphorescence lifetime) is dependent on the concentration of oxygen, but not upon the concentration of the probe, the excitation light intensity or signal collection efficiency. Data represent the average of multiple measurements that were taken at different locations on the tumor.

Treated or control H460-mCherry tumors (2–5 pooled/condition) were excised and studied for the presence of treatment-induced apoptosis (labeled by Annexin V) and necrosis (labeled by DAPI). The staining protocol is detailed in Supplementary Materials and Methods. Samples were analyzed on a LSRII Flow Cytometer using Diva software (Becton Dickinson). Data were analyzed in FlowJo (TreeStar).

To detect BPD, animals were imaged 3 hours after BPD injection. For perfusion imaging, IRDye800 PEG (LI-COR) was administered i.v. 15 to 30 minutes before imaging at 1, 4, and 18 hours after PDT. Imaging was performed on a LI-COR Pearl Impulse using white light, 700 nm, and 800 nm to visualize anatomy, BPD, and IRDye800, respectively. For analysis, similarly sized tumor and flank-identifying regions of interest (ROI) were drawn on the white light images and applied to the near-infrared images. Data are expressed as average fluorescence intensity (FI)/unit area, and for IRDye800 images, data are normalized to the flank in order to provide for individual standardization of dye circulation. For scab measurements, ROIs were drawn around scab-containing areas in the white light images, and data are expressed as the number of pixels in the scab (or the pre-PDT tumor).

Tumor-localized concentrations of human VEGF were assayed by ELISA (R&D Systems). Excised tumors were immediately frozen in a slurry of dry ice and 70% ethanol, then homogenized on dry ice and resuspended in cold 1× PBS. Vortexed samples were freeze-thawed (three times), centrifuged, and protein concentrations of cleared lysates were detected using BCA reagent (Thermo Scientific). The manufacturer's protocols were subsequently followed. Plates were analyzed using SoftMax Pro software (Molecular Devices).

EGFR expression in lysed solutions of SVEC, H460, and A549 cells was confirmed by Western blotting. Probing with anti-EGFR antibody (Cell Signaling Technology) was performed overnight on equivalent amounts of protein electrophoresed on BIS–Tris NuPage gels (Life Technologies) and transferred to nitrocellulose membranes. Anti-β-actin antibody was used as a loading control. Details on the staining protocol are provided in the Supplementary Materials and Methods.

Viability/proliferation was assessed in cells that were either left untreated or exposed to erlotinib-alone, light-alone, BPD-alone, DMSO-alone, PDT-alone, or erlotinib/PDT. Specific to the condition studied, cells were pretreated with 4 μmol/L erlotinib 2 days before PDT. The following day (1 day pre-PDT), cells were reseeded into black-walled 96-well plates (4 wells/condition), and incubated overnight. On the day of PDT, BPD (100 ng/mL) and/or erlotinib (4 μmol/L) were added for a 3-hour incubation, cells were illuminated (5 mW/cm², 0.5 J/cm²), and replacement medium was added that contained a low dose (1 μmol/L) of erlotinib (accounting for cell exposure to erlotinib in the short term after PDT in vivo). At 20 hours after PDT, the medium was replaced with alamarBlue-containing medium (cells were not reseeded) and the plate was read every hour up to 25 hours after PDT. A 5-hour incubation in alamarBlue reagent produced a strong but unsaturated signal, thus viability was assessed at the 25-hour time point per the manufacturer's protocol. Data are expressed as viability relative to untreated controls.

Cells were incubated with 4 μmol/L erlotinib at 2 days, 1 day, and 3 hours before their extraction. BPD (100 ng/mL) was included with the final addition of erlotinib. After 3 hours, cells were washed 2× in Cell Rinse (0.1 mol/L NaCl, 5 mmol/L KCl, 1 mmol/L NaH₂PO₄, 6 mmol/L glucose, 20 mmol/L HEPES, pH 7.2), extracted in 1% SDS, passed through a 30 g needle two to three times to shear the DNA, and stored frozen (−20°C). For analysis, samples were heated (50°C), diluted in PBS, and read on a Fluoromax-4 spectrofluorometer (Horiba Scientific) with excitation at 405 nm. Peak emission between 688 and 693 nm was determined and BPD concentration was calculated by comparison with a standard curve. Protein concentrations were determined using a BCA assay (Pierce) and data are expressed as BPD uptake per unit protein.

Animals received either single- or triple-dose erlotinib treatment. Three hours after the final erlotinib dose was administered, anesthetized animals underwent a terminal cardiac puncture. Blood coagulated at room temperature (∼20 minutes), and plasma was collected by centrifugation and frozen (−20°C). Bilirubin content was analyzed by Antech Diagnostics.

Following the protocol used in the alamarBlue assay, cells that were untreated, as well as those that were erlotinib and/or PDT-treated, were plated at defined concentrations and incubated to permit colony formation (∼7 days). Tests of a single “dose” of erlotinib were performed as a 3 hour erlotinib exposure before PDT. A slightly higher light dose (0.8 J/cm²) was used for clonogenics due to the more protracted time course of this assay. Surviving fractions were calculated relative to the plating efficacy of cells exposed to neither erlotinib nor PDT.

Means and standard errors are reported in figures. The Student t tests (JMP, SAS Institute, Inc.) were used to assess differences in means between groups, with paired tests used where noted. Tumor response data were described using the Kaplan–Meier method, yielding median regrowth times, and 95% confidence intervals (95% CI) noting that in some cases the upper bound was undetermined. We fit a Cox model to the data and determined differences in median survival between groups based on a Wald test. A post-hoc test for trend was used to determine whether median survival increased as a function of erlotinib dose (zero, one, or three doses). Hypothesis tests were one-sided for the tumor response data, as there was no scientific interest in determining whether treatment reduced response, and two-sided for all other hypothesis tests. The type I error rate was 0.05.

We first considered the therapeutic effect of erlotinib administration before PDT. At 2 days, 1 day, and 3 hours before light delivery, erlotinib was administered to mice bearing H460 tumors (Fig. 1A). All animals treated with erlotinib regrew their tumors in less than 1 day (i.e., no tumor regression occurred). In mice treated with PDT, median time-to-regrowth was 11 days (95% CI, 5–29 days), and impressively, pretreatment with erlotinib increased the median time-to-regrowth to over 90 days after PDT (P < 0.001 compared with PDT). This was accompanied by an increase in 90-day complete response rate from 16% (PDT) to 63% (erlotinib/PDT). Erlotinib similarly improved response in mice bearing A549 tumors (Fig. 1B). Pretreatment with erlotinib increased median time-to-regrowth to 69 days versus 8 days in animals that received PDT (P = 0.033). This was reflected in an increase in 90-day complete response rate from 15% (PDT) to 51% (erlotinib/PDT).

These findings identify exceptional improvements in therapeutic outcome as a consequence of adding erlotinib to PDT. The results were far superior to the benefit that was expected from our previous in vitro work in which erlotinib augmented the cytotoxic effect of PDT by <1 log of kill (12). As such, we considered the potential contribution of in vivo–specific factors to tumor response with erlotinib/PDT. Others have reported short duration treatment with erlotinib to increase tumor oxygenation (13), which could improve the in vivo effectiveness of PDT relative to in vitro findings (for which oxygen limitation is not an issue). To test for this, tumor oxygenation was measured after the administration of erlotinib. Compared with baseline values in each tumor, no appreciable increases in oxygenation were observed after erlotinib administration (Supplementary Fig. S1).

Flow cytometry was performed to study the nature and time course of treatment-induced cytotoxicity (Supplementary Fig. S2). In agreement with tumor response results, erlotinib was minimally cytotoxic to H460 tumors. When used as a single agent, erlotinib induced only small, insignificant increases in apoptosis (augmented by the addition of PDT) that did not affect overall viability. In the absence of PDT, erlotinib did not induce necrosis. Moreover, these data revealed erlotinib-mediated increases in PDT cytotoxicity at delayed (18 hours), but not early (1 hours) time points after treatment. This time course is consistent with PDT cytotoxicity secondary to in vivo microenvironmental effects, such as vascular damage, and led us to next evaluate the contribution of vascular effects to the erlotinib-enhanced PDT response.

Near-infrared in vivo imaging of tumor perfusion was used to study the antivascular effects of adding erlotinib to the pre-PDT regimen. In vivo imaging incorporated signal that originated mostly from the tumor vasculature, but unavoidably included signal from overlying skin, and, to some extent, underlying tissue. Thus, we compared in vivo to ex vivo imaging of tumors in order to confirm that in vivo imaging was representative of the tumor-specific signal (obtained by ex vivo imaging). The data obtained from ex vivo and in vivo images were extremely well correlated (Supplementary Fig. S3), validating use of in vivo imaging to study tumor vascular response.

By in vivo imaging, we could identify localized increases in dye accumulation, indicative of a hyper-permeable state that is characteristic after PDT (14), as well as dye exclusion that demonstrates impairment of tumor blood flow. PDT of H460 tumors produced a hyper-permeable state at 1 hour that was followed at 4 and 18 hours by the progressive development of dye-excluding vascular damage (Fig. 2). In both H460 and A549 tumors (Fig. 2, inset), there was significantly greater dye exclusion at 18 hours after PDT when animals were pretreated with erlotinib.

EGFR inhibition can decrease the expression of vascular endothelial growth factor (VEGF), a protein critical for reestablishment of damaged vasculature (15). Like other cancer therapies, PDT can induce increases in VEGF that contribute to tumor recurrence after treatment (16–18). Therefore, if erlotinib attenuates PDT-induced increases in VEGF, we might expect these decreases to contribute to the stronger vascular response in tumors pretreated with erlotinib. Indeed, PDT produced large increases in tumor concentrations of H460-produced (human) VEGF (Fig. 3A). Pretreatment with erlotinib significantly abrogated this effect and VEGF levels were barely elevated after PDT. PDT of A549 tumors led to even larger increases in human VEGF that were reduced nearly 2-fold when erlotinib was given before PDT (Supplementary Fig. S4).

To assess if abrogation of VEGF in the early time course after PDT is responsible for augmentation of PDT response by erlotinib, we investigated whether inhibition of extracellular VEGF resulted in similar improvements in PDT response. To accomplish this, we substituted a clinically approved antibody that specifically targets VEGF, bevacizumab, for erlotinib. These studies followed the rationale that if reduction of extracellular VEGF was responsible for mediating vascular shutdown after treatment with erlotinib/PDT, then bevacizumab/PDT would result in similarly robust vascular shutdown. The H460 tumor model was used because in this model the administration of erlotinib provided complete abrogation of PDT-induced increases in VEGF. Pre-PDT administration of bevacizumab reduced VEGF levels in H460 tumors after PDT below those observed in untreated controls (Fig. 3A). However, despite the dramatic reduction of VEGF in H460 tumors as a result of the pre-PDT addition of bevacizumab, no enhanced vascular damage was observed after PDT (Fig. 3B). In contrast to improvements in tumor response after treatment with erlotinib/PDT (see Fig. 1A), pretreatment with bevacizumab yielded no improvement and all tumors regrew within 26 days of illumination (median 10 days; data not shown).

To further consider the mechanisms of erlotinib-mediated augmentation of PDT vascular damage, the direct cytotoxic action of erlotinib/PDT on endothelial cells was studied. The viability of mouse endothelial cells (SVEC) was compared after treatment with erlotinib, PDT, or the combination. Prior to initiation of these experiments, EGFR expression by SVEC was confirmed by Western blot analyses and was detected at similar levels as were observed in the NSCLC tumor lines studied (Supplementary Fig. S5). SVEC viability was significantly reduced by PDT and erlotinib, as well as by the erlotinib/PDT combination (Fig. 4A). Moreover, erlotinib pretreatment before PDT significantly increased the cytotoxic effect of PDT.

A mechanism by which erlotinib could increase PDT cytotoxicity is through the capability of receptor tyrosine kinase inhibitors (TKI) to augment photosensitizer retention. This is mediated via their inhibition of drug transport by ABCG2 (ATP-binding cassette transporter G2). TKIs act through this mechanism to increase tumor levels of several photosensitizers, although BPD is typically less affected (19). Indeed, we previously found no effect of erlotinib on BPD uptake in H460 cells (12). In this study of endothelial cells, erlotinib did lead to a small but significant increase in BPD levels (Fig. 4B) that consisted of a change from 0.76 ± 0.06 nmoles BPD/μg protein in BPD-exposed cells to 0.97 ± 0.07 in erlotinib/BPD-exposed cells. Although small in magnitude, these data did prompt us to question in vivo photosensitizer retention after erlotinib pretreatment.

To elucidate whether erlotinib-induced increases in photosensitizer uptake were of significance in vivo, the fluorescence intensity of BPD in H460 tumors was directly measured by in vivo optical imaging at 700 nm. Fluorescence intensity at 700 nm increased by approximately 50% when animals were pretreated with erlotinib. With erlotinib pretreatment, BPD levels were also elevated in the flanks of the animals, indicating a more systemic impact of erlotinib on photosensitizer uptake (Fig. 5A). Moreover, similar elevations were detected in A549 tumors (Supplementary Fig. S6).

The contribution of erlotinib-induced increases in BPD uptake to therapeutic response was tested using an alternative schedule of erlotinib dosing. We found that in contrast to the triple-dosing schedule, a single dose of erlotinib given immediately before BPD administration did not increase BPD uptake (Fig. 5B). This finding may reflect the known induction of hepatic side effects by erlotinib (common among TKIs; refs. 20, 21) because patients with mild hepatic dysfunction exhibit increases in the plasma half-life of BPD (22). Indeed, we found that bilirubin, a breakdown product of heme catabolism whose elevation is associated with liver dysfunction, was five times higher in the plasma of mice given three doses of erlotinib (1.03 mg/dL ± 0.17) than in untreated controls (0.21 mg/dL ± 0.03). In contrast, a single dose of erlotinib administered 3 hours before PDT (before BPD administration) caused only a small increase in bilirubin concentration (Fig. 5C).

Thus, a single dose of erlotinib did not alter tumor BPD levels. Yet, importantly, a single dose of erlotinib does increase PDT-induced cytotoxicity to SVEC, indicating that a single dose is adequate at a molecular level to augment cell death from PDT (Supplementary Fig. S7). From these results, we hypothesized that treatment with a single erlotinib dose would produce little improvement in tumor response compared with PDT-alone if erlotinib-induced increases in BPD uptake were the sole mechanism of enhanced tumor response. Conversely, if the causative mechanisms included other factors in addition to enhanced BPD uptake, then the single-dose erlotinib schedule should yield a tumor response that is intermediate to the effects of treatment with PDT and with erlotinib/PDT on a triple-dose schedule. Our data support the second hypothesis. A log-rank test for trend indicated an increasing median regrowth time for triple-dose erlotinib/PDT, followed by single-dose erlotinib/PDT and lastly PDT-alone (P = 0.009; Table 1). Specifically, treatment with PDT after a single dose of erlotinib yielded a median regrowth time of 21 days (95% CI of 10 days and undetermined), a value that was significantly lower than regrowth time after a triple-dose erlotinib schedule (median of >90 days; P = 0.042) and higher than regrowth time after PDT-alone (median of 11 days; P = 0.053). Because response to erlotinib/PDT on a single-dose schedule is independent of an elevation in BPD levels, these data show that the therapeutic action of erlotinib/PDT includes factors other than photosensitizer uptake, such as erlotinib-induced augmentation of PDT vascular response.

Because erlotinib-induced enhancement of photosensitizer levels partly contributes to tumor response, and because erlotinib led to general increases in BPD levels, we lastly considered the selectivity of PDT when preceded by erlotinib dosing. Irrespective of whether the animal received erlotinib, scabs that developed 18 to 30 hours after PDT encompassed areas that were smaller than their pre-PDT tumor sizes (Fig. 6). Thus, the addition of erlotinib to PDT did not promote a nonselective reaction. However, the PDT-induced scabs on animals that received erlotinib/PDT were larger than those on animals that received PDT. Accordingly, they involved a larger area of the treated tumor (without nonselectively extending beyond it), which corroborates the benefit that erlotinib pretreatment provided to tumor response to PDT.

Combinations of molecular targeting agents with chemotherapy or radiation are under investigation, especially in treatment of NSCLC (23, 24), following the premise that addition of these agents to standard therapies could provide a multifaceted treatment approach (25). Similar to other therapies, there exists a benefit to combining PDT with molecular targeting (26, 27). EGFR inhibition using receptor-targeting antibodies and small-molecule inhibitors can increase the cytotoxicity of PDT against tumor cells in vitro and reduce tumor burden after PDT in vivo (7–10, 28). Furthermore, molecular targeting can be used for photosensitizer delivery through the integration of photosensitizers into antibody (e.g., cetuximab)-coated nanoparticles or by coupling of antibodies or molecular targeting agents (e.g., erlotinib) directly to a photosensitizer (29–34). In this way, molecular targeting is exploited to both increase the selectivity of PDT and augment its cytotoxicity. These studies, together with our current work, demonstrate great potential in combining PDT with molecular targeting agents to improve therapeutic outcomes. They emphasize a need to discern the mechanisms that drive improvements in PDT response, toward the goal of informing translation into clinical protocols.

When erlotinib was used as a standalone treatment, it completely failed to affect tumor growth compared with control; indeed, failure rates were identical in the two groups. In contrast, when combined with PDT, erlotinib significantly improved therapeutic outcomes in H460 and A549 tumors, and the effect was larger than PDT alone. The combined effect was, thus, by definition synergistic in nature. Both H460 and A459 cells are relatively resistant to erlotinib (35), although it is reported that a 3-week dosing schedule can produce tumor growth delay in vivo (36). This schedule is more prolonged than the triple-dose pre-PDT schedule that we used, which may account for the standalone effect of erlotinib. Effectiveness can increase at higher doses, but 100 mg/kg is the maximum-tolerated dose (36) and we found it to be associated with morbidities. Thus, the present studies were conducted at an erlotinib dose of 50 mg/kg, which was well tolerated by the animals.

One mechanism by which erlotinib enhances tumor response to PDT appears to be through the augmentation of PDT vascular effects. Vascular damage contributes to PDT response with many treatment regimens (37, 38). In BPD-PDT, acute vascular shutdown occurs when treatment is performed at short drug-light intervals (e.g., 15 minutes; refs. 38, 39). Yet, longer drug–light intervals are associated with venodilation (40), followed by reduced perfusion at delayed time points (41, 42). For example, a 50% reduction in tumor perfusion is reported at 6 hours after BPD-PDT with a 3-hour drug–light interval (41). On the basis of in vivo imaging of a perfusion marker, we find no decrease in tumor perfusion at 4 hours after PDT and an approximately 35% decrease at 18 hours, compared with controls. Erlotinib pretreatment augmented vascular effects, producing a 65% reduction in perfusion at 18 hours after PDT. Enhancement of PDT-induced vascular effects by erlotinib may be a result of endothelial cell damage. EGFR inhibition can reduce the proliferation (43) and promote apoptosis (44) of endothelial cells in tumors. We found that erlotinib reduced the viability of endothelial cells after PDT compared with PDT or EGFR inhibition as monotherapies. These results suggest that EGFR inhibition can increase the cytotoxicity of PDT to endothelial cells, thereby contributing to the vascular-damaging effects of BPD-PDT. Thus, our findings uniquely identify the tumor vasculature as an important target when erlotinib is administered in the pre-PDT time course.

Another mechanism for erlotinib's effect on PDT response stems from observations that TKIs can improve photosensitizer uptake (19). Indeed, we found that erlotinib augmentation of PDT cytotoxicity to endothelial cells was accompanied by small but significant increases in BPD uptake. A triple-dose erlotinib schedule also increased in vivo uptake of BPD, but BPD uptake was minimally affected when the dosing schedule was shorted to a single erlotinib dose. When combined with in vivo PDT, a single erlotinib dose was not as effective as the triple-dose schedule, but still provided for better outcomes than if PDT was performed alone. Single doses of erlotinib also remained effective in augmenting PDT cytotoxicity to endothelial cells. Thus, these data suggest that changes in photosensitizer uptake are one component, but not the sole contributor to erlotinib's effect in PDT.

Other potential mechanisms were also explored for the therapeutic benefit of erlotinib to PDT. For example, tumor blood flow can improve after TKI therapy (13, 45), which could potentially benefit oxygenation and PDT response. However, erlotinib did not improve oxygenation of H460 tumors in our studies, nor in the models recently published by others (46). EGFR inhibition can also reduce tumor VEGF levels (47), and, in agreement, we found erlotinib pretreatment to reduce PDT-induced acute increases in VEGF. Use of antiangiogenic agents after PDT to decrease VEGF production is associated with better long-term tumor response (16, 26, 48). However, typical combinations of antiangiogenics with PDT include prolonged, post-PDT administration of these drugs, not the exclusive pre-PDT administration of drug that was used in this study. To inform if therapeutic outcome was aided by decreases in extracellular VEGF in the early post-PDT time course, we used pre-PDT administration of bevacizumab. Results demonstrated that tumor VEGF levels were reduced by bevacizumab in the 18 hours after PDT, but without benefit to long-term PDT response. These results are not in opposition to favorable reports of combining antiangiogenics with PDT due to the pre-PDT and short time course of drug delivery in our studies. However, they do suggest that decreases in extracellular VEGF in the immediate time frame of PDT cannot account for the therapeutic benefit of pretreatment with erlotinib.

Toward the goal of clinical translation, the short pre-PDT time course of erlotinib administration in this study provides a number of benefits. First, a pre-PDT schedule of erlotinib could be adapted to clinical trials of intraoperative PDT, where it is difficult to administer oral agents to postoperative patients who are intubated. In addition, compared with typical long-term erlotinib dosing, an abbreviated schedule may avoid erlotinib side effects of skin rash and diarrhea, which typically develop in the first month (49). Limited use of erlotinib could further avoid the development of acquired resistance to TKIs that is prevalent with long-term drug use (50). Consequently, an abbreviated dosing schedule might minimize the incidence of adverse events and prevent acquired resistance, thereby improving patient quality of life compared with more typical schedules of erlotinib administration.

In conclusion, these studies uniquely investigate the pre-PDT administration of erlotinib in a clinically relevant regimen for adding molecular targeting to intraoperative PDT. Given the prevalence of tumor resistance to molecular targeting drugs, studies were conducted in erlotinib-resistant tumor models. Nevertheless, we find that short-term pre-PDT administration of erlotinib produces large improvements in PDT outcome. Multiple mechanisms are responsible for this effect, including erlotinib sensitization of the tumor vasculature to PDT and an erlotinib-induced increase in tumor accumulation of BPD. Given the multiple actions of erlotinib on PDT response and the large magnitude of its effect on this response, these findings suggest a significant potential for PDT with pre-administration of erlotinib in the treatment of NSCLC and other types of malignancies.

No potential conflicts of interest were disclosed.

Conception and design: S.M. Gallagher-Colombo, K.A. Cengel, T.M. Busch

Development of methodology: S.M. Gallagher-Colombo, K.A. Cengel, S.A. Vinogradov, T.M. Busch

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.M. Gallagher-Colombo, J. Miller, K.A. Cengel, T.M. Busch

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.M. Gallagher-Colombo, K.A. Cengel, M.E. Putt, T.M. Busch

Writing, review, and/or revision of the manuscript: S.M. Gallagher-Colombo, K.A. Cengel, M.E. Putt, T.M. Busch

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.M. Gallagher-Colombo, K.A. Cengel, T.M. Busch

Study supervision: S.M. Gallagher-Colombo, T.M. Busch

The authors acknowledge the assistance of Drs. Julie Czupryna and Jim Delikatny (Small Animal Imaging Facility), Sarah Hagan (spectrofluorometric studies), and Shirron Carter (manuscript preparation).

This study was supported by the NIH (CA-129554, CA-087971, CA-085831, and T32-CA-009677).

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.