Osimertinib, an Irreversible Next-Generation EGFR Tyrosine Kinase Inhibitor, Exerts Antitumor Activity in Various Preclinical NSCLC Models Harboring the Uncommon EGFR Mutations G719X or L861Q or S768I

The common activating mutations ex19del and L858R account for between 80% and 90% of the activating mutations in EGFR associated with non–small cell lung cancer (NSCLC). The remaining 10% to 20% are referred to as uncommon activating mutations and are a diverse group of mutations in exons 18 to 21 within the kinase domain of the EGFR gene (1, 2). Excluding those found as insertion mutations in exon 20, the uncommon mutations involving codons G719, S768, and L861 are the most prevalent (3–7). The G719X point mutations, where “X” represents any of the possible substitutions, account for collectively 2% to 5% of EGFR-mutated tumors (COSMIC 2018), and result in amino acid substitutions within exon 18 of EGFR at codon position 719 glycine (G): primarily glycine to alanine (G719A), glycine to cysteine (G719C), and glycine to serine (G719S). Mutations affecting G719 can occur alone or in compound mutations with other uncommon mutations, such as S768I or L861Q (3–7). The S768I point mutation accounts for 1.5% to 3% of untreated EGFR-mutated tumors (3–7) and results in amino acid substitution within exon 20 in EGFR at codon position 768, from a serine (S) to an isoleucine (I). The L861Q point mutation accounts for approximately 2% to 3% of EGFR-mutated tumors (3–7) and results in amino acid substitution within exon 21 in EGFR at codon position 861, from a leucine (L) to a glutamine (Q). These uncommon mutations as well as numerous, less prevalent mutations have been detected in NSCLC, and many of these have been reported as both activating and sensitive to EGFR tyrosine kinase inhibitors (TKI) including osimertinib (8).

However, the clinical effectiveness of EGFR-TKI treatment in patients with these uncommon mutations remains unclear because few clinical trials have enrolled patients with G719X, L861Q, or S768I (3, 9–13). To date, the largest clinical study specially designed to evaluate the responses of gefitinib and erlotinib treatment in NSCLC patients with G719X, L861Q, and S768I was composed of 78, 57, and 7 patients, respectively (9). Patients with EGFR Ex19del or L858R point mutations who had been treated with EGFR-TKIs during the same period were selected as the control group. The trial revealed that patients with exon 19 deletions had the most favorable progression-free survival (PFS; median, 13.5 months) and an overall response rate (ORR) of 65.3%, followed by those with L858R (median, 10.4 months; ORR, 67.5%), L861Q (median, 8.1 months; ORR, 39.6%), G719X (median, 6.3 months; ORR, 36.8%), and S768I (ORR, 33.3%) mutations (9). Additionally, compound mutations in 19 patients (G719X;L861Q in 9 patients and G719X;S768I in 10 patients) were evaluated with ORR of 88.9% and 50.0%, respectively.

Yang and colleagues retrospectively reviewed patients with the G719X, L861Q, or S768I mutations alone or in combination who were treated with the second-generation EGFR-TKI, afatinib, in the LUX-LUNG trial series and found that the ORR was 77.8%, 56%, and 100%. The PFS was 13.8, 8.2, and 14.7 months, respectively (14).

Osimertinib is an oral, third-generation, irreversible EGFR-TKI that selectively inhibits both EGFR-TKI–sensitizing and EGFR T790M–resistance mutations with lower activity against wild-type EGFR and has demonstrated efficacy in NSCLC CNS metastases (15–19). When used in the first-line setting to treat patients with NSCLC harboring the canonical Ex19del or L858R mutations, osimertinib has shown more favorable treatment outcomes than first-generation and second-generation EGFR-TKIs including in patients with CNS metastases (20). Recently, a paper by Ahn and colleagues shows that osimertinib demonstrated favorable activity with manageable toxicity in patients with NSCLC harboring uncommon EGFR mutations (10). Thirty-six patients with uncommon mutations were evaluated, and the objective response rate was 50% with a median PFS of 8.2 months.

Using in vitro cellular EGFR phosphorylation assays, we showed that osimertinib was potent against a range of uncommon mutant variants observed clinically of EGFR involving codons G719, S768, L861, and L747. Using available patient-derived xenografts (PDX) and cell lines derived from two of these PDXs that harbor G719X mutation, we have evaluated in vitro and in vivo the preclinical activity of osimertinib. We report osimertinib inhibits signaling pathways and cellular growth in G719X-mutant cell lines in vitro and demonstrates sustained tumor growth inhibition of PDX harboring G719X mutation in combination with L861Q or S768I.

This work demonstrates that osimertinib is active against a range of uncommon mutant variants of EGFR and offers potential to improve treatment options for patients with NSCLC whose tumors harbor an uncommon EGFR mutation.

Cos-7 cells were obtained from European Collection of Authenticated Cell Cultures (ECACC). Cos-7 cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM, Sigma-Aldrich) supplemented with 10% fetal calf serum (FCS; PAA) and 1% Glutamax (Life Technologies). Cos-7 cells were authenticated at AstraZeneca cell banking using DNA fingerprinting short tandem repeat assays and confirmed to be free of bacterial and viral contaminations by IDEXX. All cell lines were used within 15 passages and for a period of less than 6 months. The patient-derived cell lines YU-1092 and YU-1099 were derived from malignant effusions of NSCLC patients (21). Both cells were initially cultured on collagen-coated plates in a serum-free defined medium (ACL-4) supplemented with 5% FBS. The cells maintained the driver oncogenes that were observed in the patients. When cells were enriched in an epithelial cell adhesion molecule (EpCAM)-positive cell population with purity of over 95%, cells were subjected to further assays. All patient samples were obtained after written informed consent from the patients using protocols approved by the institutional review board.

Osimertinib, AZ5104, and afatinib have been synthetized by AstraZeneca. The synthesis and structures of osimertinib and AZ5104 have been previously reported as compounds 8 and 27 in ref. 22.

EGFR phosphorylation was measured using a modified Cisbio Pan phospho-EGFR Cellular Assay Kit. All Cos-7 cells were revived in culture flasks for 16 hours in DMEM supplemented with 10% FCS. Cos-7 cells were detached using TrypLE Express Enzyme (Life Technologies) from culture flasks and resuspended, 5 μL of cell suspension at densities between 600 and 1,400 cells per well was dispensed into a Greiner low volume 384 proxi plates predosed with titrations of test compound, and incubated for 2 hours at room temperature. The cells were stimulated for the final 10 minutes with 200 ng/mL of EGF. Following the 2-hour incubation, 2 μL of the XL665 and Cryptate-labeled antibodies diluted in lysis buffer were added to the cells and incubated for 2 hours at room temperature. Plates were then read on Pherastar with a HTRF module. IC₅₀ values were determined following 2-hour incubation with compound titrations. Data were exported to Genedata Screener software package (Genedata) to generate sigmoidal dose−response curves, and an IC₅₀ value was generated by determining the compound concentration at which there was a 50% inhibition in signal.

Bim, EGFR, p-EGFR (Y1068), AKT, pAKT (S473), ERK, pERK (T202/Y204), S6, pS6 (S240/244), and horseradish peroxidase (HRP)–conjugated secondary antibodies were purchased from Cell Signaling Technology. Actin was obtained from Merck Millipore. HRP-dependent luminescence was developed with SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific) and detected with LAS-4000 lumino-image analyzer system (Fujifilm).

Cells were seeded at a density of 5,000 cells per well on 6-well plates. Cells were incubated overnight and exposed to the indicated drugs for 14 days. Medium containing drugs were replenished every 3 days. Then, cells were stained with crystal violet and counted using ImageJ software.

All animal studies were conducted in accordance with UK Home Office legislation, the Animal Scientific Procedures Act 1986, as well as the AstraZeneca Global Bioethics policy. Randomization of animals onto study was based on initial tumor volumes to ensure equal distribution across groups. A power analysis was performed whereby group sizes were calculated to enable statistically robust detection of tumor growth inhibition (≥ 5 per group) or pharmacodynamics endpoint (≥ 4 per group).

For the LC-F-29 (Xentech), the CTG-1082 and CTG-2534 (Champions Oncology) and LU1901 (Crown Bioscience) PDX models, tumor fragments from donor mice inoculated with primary human lung cancer tissues were harvested and inoculated subcutaneously into the flank of nude female mice. Tumor growth was monitored twice weekly by bilateral caliper measurements and tumor volume calculated using the formula TV (cm³) = [length (cm) × width (cm)²] × 0.5, where the length and the width are the longest and the shortest diameters of the tumor.

Mice were randomized into vehicle or treatment groups with approximate mean start size of 0.2 cm³. Randomization for animal studies was based on initial tumor volumes to ensure equal distribution across groups. Mice were dosed daily by oral gavage for the duration of the treatment period with vehicle, osimertinib, afatinib, or savolitinib. Tumor growth inhibition (%TGI) from the start of treatment was assessed by comparison of the geometric mean change in tumor volume for the control and treated groups using the formula: %TGI = (1 − {T_t/T₀/C_t/C₀}/1 − {C₀/C_t}) × 100, where T_t is the geomean tumor volume of treated at time t, T₀ is the geomean tumor volume of treated at time 0, C_t is the geomean tumor volume of control at time t, and C₀ is the geomean tumor volume of control at time 0. Tumor regression was calculated as the percentage reduction in tumor volume from baseline value: % regression = (1 − RTV) × 100 %, where RTV = mean relative tumor volume. Statistical significance was evaluated using a one-tailed t test.

For pharmacodynamic studies, mice were randomized at a tumor volume between 0.4 and 0.6 cm³ using the same randomization criteria as the tumor growth inhibition studies. Mice were then treated orally with a single bolus dose of either vehicle, osimertinib, or afatinib. Tumors were excised at specific time points after dosing and flash frozen in liquid nitrogen.

In all in vivo studies, osimertinib, afatinib, and savolitinib were administered via oral gavage. Osimertinib, afatinib, and savolitinib were suspended in 0.5% HPMC, 0.5% HPMC + 0.1% polysorbate 80, and carboxymethylcellulose-sodium, pH 2.1, respectively.

For ex vivo, pea-sized fragments of xenograft tissue were homogenized in the FastPrep-24 5G instrument (MP Bio) in lysis buffer supplemented with protease and phosphatase inhibitor cocktails (Sigma). Homogenates were briefly sonicated using Diagenode Bioruptor plus before 10-minute centrifugation and protein quantification with Pierce protein assay (Thermo Fisher). For all samples, equal protein amounts were loaded for SDS-PAGE using 4% to 12% gradient Bis-Tris precast gels (Novex Life Technologies), followed by transfer to nitrocellulose membranes using the iBlot2 dry transfer system (Novex Life Technologies). After blocking in 5% milk-TBST or BSA-TBST, membranes were blotted with phospho-EGFR (p-EGFR; Tyr1068; Cell Signaling Technology; 2234), total EGFR (tEGFR; Cell Signaling Technology; 2232), phospho-AKT (pAKT; Ser473; Cell Signaling Technology; 4060), total AKT (tAKT; Cell Signaling Technology; 9272), phospho-ERK p(ERK) (Thr202/Tyr204; Cell Signaling Technology; 9101), total ERK (tERK; Cell Signaling Technology; 9102), phospho-S6RP (pS6; Ser235/236, Cell Signaling Technology; 4858), total S6RP t(S6RP) (Cell Signaling Technology; 2217), or vinculin (Sigma; V9131) followed by HRP-conjugated secondary antibodies (Cell Signaling Technology; 7074 or 7076). Signals were detected with SuperSignal West Dura detection reagents (Thermo Fisher). Western blots were developed using G: Box chemi genius instrument (Syngene), and where indicated, signal was quantified in Syngene Genetools software. Protein levels were then normalized to the levels of loading control (vinculin), and treatment groups were normalized to the mean of time-matched vehicle groups. Statistical significance was evaluated using a one-way, two-sided ANOVA.

To assess the potency of osimertinib against uncommon EGFR mutations, COS7 cells were transfected with cDNAs encoding EGFR G719A/C/S, L861Q, S768I, and L747S as single or compound mutations and treated with increasing concentrations of either osimertinib, AZ5104 (a circulating metabolite of osimertinib), or afatinib to determine the apparent geomean IC₅₀ for inhibition of phosphorylation of the uncommon mutants (Table. 1). Osimertinib was able to inhibit EGFR phosphorylation of all the uncommon mutations evaluated when expressed individually or in combination with apparent geomean IC₅₀ values ranging from 4.5 to 40.7 nmol/L. Osimertinib potency was similar to AZ5104 potency (apparent geomean IC₅₀ range, 1.2–40.7) and afatinib potency (apparent geomean IC₅₀ range, 1.7–79.2; Table. 1). Notably, the potency against uncommon mutations was of the same order as potency in the same assay for inhibition of the common exon 19 deletion and L858R mutants (apparent geomean IC₅₀ are of 8.4 and 11.9 nmol/L, respectively; Table. 1).

Collectively, these data indicate that osimertinib potently inhibits the activity of multiple different uncommon EGFR mutations in vitro.

Next, we evaluated the activity of osimertinib to inhibit EGFR phosphorylation and proliferation in vitro of EGFR-mutant NSCLC patient-derived cell lines carrying either the EGFR G719C/S768I (YU-1099) or the EGFR L861Q (YU-1092) mutations. In the YU-1099, the lowest concentration of osimertinib tested was sufficient to completely inhibit EGFR phosphorylation (Fig. 1A). In the YU-1092, osimertinib concentration required to completely inhibit EGFR phosphorylation was between 30 and 100 nmol/L (Fig. 1B). Therefore, consistent with engineered COS7 cell data (Table. 1), osimertinib inhibits EGFR phosphorylation of representative uncommon EGFR mutations in more disease-relevant cell models. In both cell lines, osimertinib induced concentration-dependent inhibition of downstream signaling markers (pAkt, pERK, and pS6) and increase in the levels of the proapoptotic protein BIM (Fig. 1A and B).

We then explored how the activity against EGFR uncommon mutations translated into cell proliferation inhibition. In line with the phosphorylation data, in a proliferation assay using both the YU-1099 and YU-1092 cell lines, osimertinib showed potent inhibition of proliferation with an apparent IC₅₀ of approximately 30 nmol/L and less than 10 nmol/L, respectively (Fig. 1C–F).

To explore the in vivo activity of osimertinib, we administered the drug as monotherapy against the LC-F-29 NSCLC PDX model that carries the EGFR G719A;S768I mutation. Once-daily administration of 25 mg/kg of osimertinib (a dose that approximates to the clinically approved 80-mg dose) induced significant tumor growth regression in the LC-F-29 PDX model ((>100) 45%, P < 0.001 at day 14) when compared with the control group (Fig. 2A). A dose of 7.5 mg/kg afatinib, which represents the 40-mg clinical starting dose of afatinib, induces a modest tumor growth inhibition (58%, P < 0.01 at day 14) when compared with the control group (Fig. 2A). In a repeat of the previous experiment, in the LC-F-29 PDX model, osimertinib induces a very similar tumor regression (>100; 81%, P < 0.001 at day 14) when compared with the control group (Supplementary Fig. S1A). Waterfall plots (Fig. 2B; Supplementary Fig. S1B) for each of the individual mice treated on these studies demonstrate that all mice experienced reduction in their tumor volumes following osimertinib treatment, whereas all the mice in the afatinib-treated cohort had tumor growth by day 14 (Fig. 2B). Both compounds were well tolerated and minimal body weight loss (less than 10% of starting body weight) was observed (Fig. 2C; Supplementary Fig. S1C).

To explore the relationship between efficacy and target modulation, mice bearing LC-F-29 PDX tumors were treated with a single dose of either vehicle, osimertinib, or afatinib, and tumors were harvested 1, 6, 16, and 24 hours later. Comprehensive EGFR pathway biomarkers were used to assess the impact of EGFR-TKI within the tumor tissue (Fig. 3). Pharmacodynamics effects were confirmed by assessing phospho-EGFR inhibition by Western blot (Fig. 3A and 3E; Supplementary Fig. S2A), and downstream signaling pathway inhibition by assessing phospho-ERK, phospho-S6, and phospho-AKT inhibition by immunoblotting (Fig. 3B–E; Supplementary Fig. S2B–S2D). Although the pharmacokinetic half-life of osimertinib in mice is only approximately 3 hours (15), phospho-EGFR levels remained significantly reduced even 24 hours after dose (Fig. 3A and E; Supplementary Fig. S2A), consistent with the expected irreversible mode of action. Importantly, downstream signaling markers also showed significant inhibition. All pharmacodynamic biomarkers assessed showed more transient modulation by afatinib than osimertinib, which is consistent with the differences in antitumor efficacy observed in this model (Fig. 3B–D; Supplementary Fig. S2B–S2D).

These data demonstrate that osimertinib can achieve robust pathway inhibition in vivo of EGFR-mutant NSCLC PDX carrying the EGFR G719A;S768I that is associated with tumor growth inhibition.

To explore further the in vivo activity of osimertinib against G719X mutation in the presence of co-occurring mutations, we administered osimertinib as a monotherapy against the CTG-1082 and CTG-2543 NSCLC PDX models that carry the EGFR G719A;L861Q or the EGFR G719C;S768I compound mutations, respectively. Once-daily administration of 25 mg/kg of osimertinib induced significant tumor growth inhibition (87%, P < 0.001 at day 14) and tumor regression (>100; 58%, P < 0.001 at day 14) when compared with the control group in the CTG-1082 and CTG-2543, respectively (Fig. 4A and B). Waterfall plots for each of the individual mice treated with osimertinib demonstrate that in the CTG-1082 PDX model, half of the mice show stable disease and half of the mice had modest tumor growth (Fig. 4C), whereas in the CTG-2534 model, all the mice experienced profound reduction in their tumor volumes by day 14 (Fig. 4D).

Osimertinib was well tolerated, and minimal body weight loss (less than 10% of starting body weight) was observed compared with predose starting body weight (Fig. 4E and F).

Collectively, these results demonstrate that osimertinib is active in vivo across PDX models, which harbor the G719X mutation even when co-occurring with either the EGFR L861Q or EGFR S768I and further support the potential efficacy of osimertinib in patients harboring various EGFR uncommon mutant tumors.

LU1901, which harbors the G719A EGFR mutation, was resistant to osimertinib treatment (Fig. 5A; Supplementary Fig. S3). Genomic analysis revealed that the model also harbors an amplification of MET, a well-established resistance mechanism known to bypass EGFR inhibition and impart resistance to EGFR-TKIs, including osimertinib (23). To validate the mechanism of resistance, we administered savolitinib, a type 1 MET inhibitor (24), alone or in combination with osimertinib. Once-daily administration of 25 mg/kg of savolitinib alone or in combination with osimertinib induced profound tumor growth regression (>100; 95%, P < 0.001 at day 14; >100, 100%, P < 0.001 at day 14) when compared with the control group, respectively (Fig. 5A). Waterfall plots for each of the individual mice demonstrate that in the LU1901 PDX model, 3 of 5 mice and 5 of 5 mice had no measurable tumor growth when treated with savolitinib alone or with savolitinib in combination with osimertinib, respectively, at day 14 (Fig. 5B), whereas all the mice in the osimertinib-treated cohort had tumor growth similar to control-treated cohort by day 14 (Fig. 5B). Collectively, these results demonstrate that the growth of the LU1901 model EGFR-mutant, MET-amplified is dependent on MET activation alone.

Both compounds were well tolerated, and minimal body weight loss was observed compared with predose starting body weight (Fig. 5C).

Despite the success of approved TKI therapies to treat NSCLC patients harboring activating mutations in EGFR, the benefit of treatment from these agents remains mostly limited to the common Ex19del and L858R mutant subtypes.

Alongside L858R and Ex19del, the uncommon EGFR mutations G719X, L861Q, and S768I account for 5% of EGFR mutations. This subgroup is less sensitive to first-generation TKI, erlotinib, and gefitinib. Patients with uncommon mutations had a significant shorter PFS and lower ORR than did patients with common mutations (PFS, 7.7 vs. 11.4 months; ORR, 41.6% vs. 66.5%; ref. 9). Retrospective review of the LUX-LUNG trials reveals that the ORR and PFS were very similar in between the uncommon patient population and the patient harboring canonical activating mutations following treatment with afatinib.

Previous reports have postulated the potential of osimertinib in the uncommon mutation setting; however, these studies were limited to using Ba/F3 engineered cell models in vitro (8, 15). We therefore wanted to take a more comprehensive approach to evaluating osimertinib across a wider range of uncommon mutations and using more disease-relevant models in vitro and in vivo. Using in vitro cell EGFR phosphorylation assays, we demonstrated the potency of osimertinib and AZ5104, its active metabolite, against a panel of single and compound, uncommon mutations; in all cases, the IC₅₀ for inhibition of EGFR phosphorylation was less than 50 nmol/L for both compounds. We identified two patient-derived tumor cell lines and four PDX models that harbor an uncommon mutation, which allows us to evaluate osimertinib activity both in vitro and in vivo. Using this approach, we were able to demonstrate in vitro that osimertinib inhibits EGFR phosphorylation and downstream signaling and translated into cell proliferation inhibition. The potency of these effects in PDX cell lines was consistent with the potency we observed in cell phosphorylation assays. In vivo, osimertinib delivered high levels of antitumor activity across PDX models at 25 mg/kg/day, a dose modeled to be approximately consistent with the 80-mg clinically approved dose for targeting the canonical Ex19 del and L858R mutations as well as the gatekeeper T790M tumors. Moreover, the level of efficacy achieved was significantly greater than clinically relevant doses of afatinib. The tumor inhibition activity of osimertinib was associated with potent activity against EGFR and downstream signaling pathways. Using this approach, we were able to demonstrate that osimertinib exerts profound tumor growth inhibition in tumor harboring uncommon mutations.

Of all the models evaluated, only the LU1901 model, EGFR-mutant and MET-amplified, was resistant to osimertinib. MET amplification is a well-established resistance mechanism known to bypass EGFR inhibition and impart resistance to EGFR-TKIs, including osimertinib (23). This model was dependent on MET activation alone as no additional benefit of osimertinib addition to savolitinib was observed over savolitinib monotherapy. To note, some recent reports have also described monotherapy MET inhibitor sensitivity in EGFR-mutant patients, suggesting that single-agent MET dependence can also occur clinically in patients (25, 26).

In parallel, to evaluate all the uncommon G719X, L861Q, and S768I mutations, single or compound mutation forms not represented in the two-patient-derived tumor cell lines and the four PDX models, we used an in vitro cellular EGFR phosphorylation assay. In the assay, osimertinib demonstrated potent inhibition against all the uncommon mutations evaluated at potencies comparable to the common activating ex19del and L858R mutants.

Recently, a paper by Ahn and colleagues shows that osimertinib demonstrated favorable activity with manageable toxicity in patients with NSCLC harboring uncommon EGFR mutations (10). Thirty-six patients with uncommon mutations were evaluated, and the objective response rate was 50% with a median PFS of 8.2 months. Although cross-trial comparisons should be performed with caution, the activity of osimertinib seems to be superior against the activating mutations. The objective response rate was 80% with a median PFS of 17.2 months (20). However, larger studies in patients harboring uncommon mutations will be required to evaluate this further.

Overall, we suggest the work presented here, in addition to the first clinical report, provides sufficient evidence to warrant additional clinical testing of osimertinib within this uncommon EGFR mutation population to establish whether osimertinib can offer improved clinical benefit in this important remaining area of unmet need as a differentiated next-generation TKI.

D.A.E. Cross reports personal fees from AstraZeneca (employer) outside the submitted work. B.C. Cho reports grants from Novartis, Bayer, AstraZeneca, MOGAM Institute, Dong-A ST, Champions Oncology, Janssen, Yuhan, Ono, Dizal Pharma, MSD, AbbVie, Medpacto, GIInnovation, Eli Lilly, and Blueprint Medicines (research funding), personal fees from Novartis, AstraZeneca, Boehringer-Ingelheim, Roche, BMS, Ono, Yuhan, Pfizer, Eli Lilly, Janssen, Takeda, MSD, Janssen, Medpacto, and Blueprint Medicines (consulting), personal fees from TheraCanVac Inc, Gencurix Inc, Bridgebio Therapeutics, KANAPH Therapeutic Inc (stock ownership), personal fees from KANAPH Therapeutic Inc (scientific advisory board), other from Daan Biotherapeutics (founder), and other from Champions Oncology (royalty) outside the submitted work. P.D. Smith reports other from AstraZeneca PLC (employee and shareholder) outside the submitted work. No potential conflicts of interest of were disclosed by the other authors.

N. Floc'h: Conceptualization, data curation, formal analysis, supervision, investigation, methodology, and writing–original draft. S. Lim: Conceptualization, data curation, formal analysis, supervision, and methodology. S. Bickerton: Resources, formal analysis, and methodology. A. Ahmed: Resources, formal analysis, and methodology. J. Orme: Resources, formal analysis, and methodology. J. Urosevic: Resources, formal analysis, and methodology. M.J. Martin: Supervision, writing–review, and editing. D.A.E. Cross: Supervision. B.C. Cho: Writing–review and editing. P.D. Smith: Writing-review and editing.

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.