⁸⁹Zr-panitumumab Combined With ¹⁸F-FDG PET Improves Detection and Staging of Head and Neck Squamous Cell Carcinoma

Fluorine-18 fluorodeoxyglucose PET combined with CT (¹⁸F-FDG PET/CT) is one of the most common imaging modalities for staging and surveillance of cancer. ¹⁸F-FDG is taken up by cells with high glucose metabolism, and as such, increased uptake is visualized in aggressive tumors as well as in normal physiologic states (e.g., brain cortex, genitourinary tract, bone marrow, palatine tonsils) and other states of increased metabolic activity (e.g., post-surgical changes, inflammation, and infection).

Although ¹⁸F-FDG PET/CT has high sensitivity and specificity for detection of cancer lesions for TNM (tumor, lymph node, metastasis) staging in head and neck cancer (sensitivity - T: 97%–100%, N: 82%–87%, M: 53%–100%; specificity - T: 94%, N: 79%–86%, M: 93%–98%; refs. 1–3), the prevalence of incidental findings in head and neck cancer continues to be high due to lack of tumor-specificity with ¹⁸F-FDG, especially in regional lymph nodes. Incidental lesions are identified by ¹⁸F-FDG PET/CT in 35–38% of patients with head and neck cancer, 12%–20% of whom undergo invasive procedures (e.g., biopsy and/or endoscopy) while the true positive rate for distant metastatic disease is only 5%–9% (4, 5). This results in a significant number of unnecessary invasive procedures, which may delay treatment initiation or may result in unnecessary treatment (6).

To improve whole body imaging in head and neck cancer, tumor-targeted radiotracers such as CD44v6 (a cell surface molecule; ref. 7), angiogenesis-related integrins (8), somatostatin receptors (9), and hypoxia markers (10) have been tested. Yet, heterogeneous expression of receptors, limited application to tumors with specific receptors, and immunogenicity impede their adoption. Head and neck squamous cell carcinoma (HNSCC) virtually universally overexpress epidermal growth factor receptor (EGFR). EGFR overexpression has been used as a target for tyrosine kinase inhibitors and monoclonal antibodies with modest anti-cancer effect (11–16). Moreover, using optical agents (e.g., fluorescently conjugated monoclonal antibodies) targeting EGFR in the intraoperative setting demonstrated its value for real-time tumor detection (17–21). With this successful targeting of EGFR in the intraoperative setting, we wanted to explore the value of a radiolabeled antibody for its ability to preoperatively identify tumor and potentially create a roadmap to further improve intraoperative identification of regional metastatic disease. To determine the potential of EGFR targeted whole body imaging for preoperative evaluation, we evaluated zirconium-89 panitumumab (⁸⁹Zr-pan) to detect HNSCC lesions in patients with head and neck cancer.

A single center, open label, non-randomized early phase clinical study was performed to evaluate panitumumab-IRDye800 (pan800) and ⁸⁹Zr-pan for dual-modality imaging in patients with HNSCC. ⁸⁹Zr-pan (IND 135573) and pan800 (IND 119474) were approved as investigational drugs by the FDA, and the study was approved by the Institutional Review Board (IRB-41878) and performed in accordance with the Declaration of Helsinki, ICH-GCP guidelines, and the United States Common Rule.

An overview of the trial workflow is presented in Fig. 1. A total of 14 adult patients with recurrent or new biopsy-proven HNSCC with any T stage and any subsite and ¹⁸F-FDG PET/CT performed as standard of care who were scheduled to undergo neck dissection were enrolled sequentially after obtaining written informed consent at Stanford Hospital and Clinics between 4/2019 and 12/2020. Patients with previous bilateral neck dissection, pregnant or breastfeeding, severe renal/heart/lung disease were excluded (detailed inclusion and exclusion criteria available upon request). Patients were infused with a 30 mg (n = 9) or 50 mg (n = 5) flat dose of pan800, followed by 1 mCi (37 MBq) of ⁸⁹Zr-pan 1–5 days prior to surgery. The molecular weight of panitumumab is 147 kDa, IRdye800 1.1 kDa. Trace amounts (<1 mg) of panitumumab were conjugated to ⁸⁹Zirconium.

All patients underwent one PET/CT scan using a SiPM-based camera with 3D and time of flight acquisition (Discovery MI, GE Healthcare, Chicago, IL, USA) 1–5 days after administration of ⁸⁹Zr-pan. Five patients were enrolled in the dosimetry sub-study and underwent a total of three PET/CT scans at 1–2 hours, 24–48 hours, and 72–96 hours after administration of the study drug. Whole body images were acquired from vertex to mid-thigh, and data were reconstructed with an ordered subset expectation maximization (OSEM) algorithm using 2 iterations and 17 subsets and were corrected for decay, attenuation, scattering, dead time, and randoms. A low-dose whole-body CT scan was acquired once for attenuation correction and organ delineation.

The ⁸⁹Zr-pan PET/CT and the ¹⁸F-FDG PET/CT scans were reviewed in a blinded manner by two board-certified nuclear medicine physicians (AI, HD) prior to patients undergoing their standard of care surgical resection.

The ¹⁸F-FDG PET/CT scans were evaluated first, followed by ⁸⁹Zr-pan PET/CT scans. Regions of interest (ROI), such as primary tumor, lymph nodes (LNs), and other organs were outlined, and the maximum standardized uptake values (SUV_max) (22) were quantified (HD) to evaluate exploratory thresholds.

Intraoperative near-infrared fluorescence imaging was performed as described by Nishio and colleagues (19) LNs identified on preoperative ⁸⁹Zr-pan PET/CT were correlated with the intraoperatively identified specimens based on size and location in the neck levels. Resected tissue specimens underwent routine histopathologic processing followed by EGFR immunohistochemistry for target validation as previously described (19).

Absorbed doses (ADs) were calculated using OLINDA/EXM software (version 1.1, Vanderbilt University, Memphis, TN) which bases the dose calculations on the MIRD model (23) in a similar manner as described by Lindenberg and colleagues (24).

Descriptive statistics are presented as means and standard deviations (SD) for normally distributed continuous characteristics, including tumor size, time from drug infusion to imaging; median and ranges for non-normally distributed continuous characteristics, including mean signal intensities (MSI), such as mean fluorescence intensities (MFI) and SUV_max, and signal-to-background ratios (SBR). Standard boxplots (median as center line, the first and third interquartile range (IQR) as hinges, the highest and lowest values within the 1.5*IQR as hinges, outlier as circle, and mean as diamond) were used to graphically represent the difference in MSIs among the different imaging modalities.

For the non-normally distributed data, the Kruskal–Wallis test was used to compare the MSIs and SBRs within each type of image (¹⁸F-FDG PET/CT, ⁸⁹Zr-pan PET/CT, pan800 imaging), and post-hoc Wilcoxon rank sum test was used to evaluate pairwise comparisons. For normally distributed data, the one-way ANOVA was used to compare mean MSIs and SBRs after evaluation of key assumptions, including extreme outliers, normality, and homogeneity of variances. To compare the SUV_max between the two PET/CT scans, Wilcoxon signed rank test was performed to evaluate the paired, non-normally distributed data. Sensitivity and specificity in detecting HNSCC among the different imaging modalities were calculated using the receiver operating characteristic (ROC) curves on a per-lesion basis. The optimal cutoff values for MSI and SBR were calculated using the Youden index, which is (sensitivity + specificity) – 1. Logistic regression was performed using ¹⁸F-FDG SUV_max and ⁸⁹Zr-pan SUV_max to predict the binary response variable of histopathology – whether cancer was present or not. For all analyses, a two-tailed P value of less than 0.05 was deemed statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001). The relationship between SUV_max and MFI was assessed using the Pearson correlation coefficient, the coefficient of determination (R²) value, and the Spearman's correlation coefficient. Repeated measures ANOVA was performed to compare mean SUV_max across different scan times of the ⁸⁹Zr-pan PET/CT after confirming the following assumptions of no significant outliers, normal distribution using the Shapiro-Wilk test and plotting QQ plots, and homogeneity of the variances using Mauchly's test. Missing SUV_max data was limited (5.1% and 7.7% for ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT, respectively) and was omitted for analysis.

Powe considerations were limited given this was an early Phase 1 study. However, given our expectation of 420 specimens (14 subjects and average 30 LNs per subject), we calculated the sample size based on prevalence of 7.5%, significance level of 0.05, 0.3 proportion of discordant pairs, 95% power to detect 0.1 difference in specificities, and greater than 71% power to detect 0.25 difference in sensitivity in detecting metastatic lymph nodes. However, we evaluated only those lesions that were detected on either ¹⁸F-FDG PET/CT or ⁸⁹Zr-pan PET/CT with their corresponding histopathology to compare the diagnostic performance of the two radiotracers. Given the potential for underpowered analysis, interpretation based on 95% confidence intervals was recommended.

The study protocol and the data generated in this study are available upon request from the corresponding author.

There were 14 patients enrolled and 17 neck dissection specimens since 3 patients underwent bilateral neck dissection (Table 1, Supplementary Fig. S1). The clinical N stage was evenly distributed between N0 (no nodal disease) and N+ (nodal disease). One patient underwent neoadjuvant immunochemotherapy with good clinical response prior to surgical resection with pathologic staging showing no evidence of residual cancer. All patients received both ⁸⁹Zr-pan and pan800. The median time between ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT was 23 days (IQR 14.8 days), and there were no clinical interventions between the two imaging studies. All patients received 1.00 ± 0.12 mCi (mean ± SD) of ⁸⁹Zr-pan followed by 30 mg or 50 mg of pan800. The mean ± SD time interval from ⁸⁹Zr-pan administration to ⁸⁹Zr-pan PET/CT was 73.5 ± 31.6 hours, and the time interval from pan800 infusion to the first fluorescent intraoperative image was 96.1 ± 37.0 hours. Overall, a total of 37 lesions were evaluated on both ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT scans and histopathology; 11/12 primary tumors were positive for disease and 10/25 regional LNs were positive for disease on histopathology. In addition, 11 incidental lesions outside the head and neck region were included for analysis, but pathology was only available for one lesion. One grade 1 adverse event of prolonged QTc (460 ms) was observed and deemed possibly related to the study on day of infusion, but this resolved in follow-up. There were no grade 2 or 3 events that were related to the study drug (Supplementary Table S1).

All lesions that were detected on ¹⁸F-FDG PET/CT and confirmed to be HNSCC on histopathology were identified on ⁸⁹Zr-pan PET/CT (Fig. 2). ¹⁸F-FDG PET/CT had higher median SUV_max compared to ⁸⁹Zr-pan PET/CT for primary tumor (11.6 vs. 3.4; P < 0.001; Wilcoxon sign rank), LN with cancer (11.6 vs. 2.4; P = 0.009; Wilcoxon sign rank), and LN without cancer (4.7 vs. 2.4; P < 0.001; Wilcoxon sign rank) with the following ranges for primary tumor, LN with cancer, LN without cancer (¹⁸F-FDG: 6.5–22.6, 7.5–23.6, 2.0–21.0; ⁸⁹Zr-pan: 0.94–10.5, 1.06–9.8, 1.4–2.4), respectively (Fig. 3B). Consistent with previously published studies using optically labeled panitumumab (19, 20, 25), both MFI and SBR in this study were able to distinguish benign tissue from tumor tissue with high sensitivity and specificity [MFI: 85.2% (95% confidence interval (CI), 72%–99%], 80.2% (95% CI, 74%–86%); SBR: 96.3% (95% CI, 89.2%–100%), 91.0% (95% CI, 87% – 95%), respectively) with cutoff values of 0.016 for MFI and 2.9 for SBR. SUV_max of the ⁸⁹Zr-pan PET/CT and MFI of the fluorescence imaging were more highly correlated (Pearson = 0.82; R² = 0.67; Spearman = 0.66) than the SUV_max of ¹⁸F-FDG PET/CT with MFI (Pearson = 0.18; R² = 0.03; Spearman = 0.09) in detecting tumor lesions (Fig. 3C).

Using the SUV_max and histopathology as the reference standard, ROC curves were evaluated for the SUV_max of the two scans for only the lesions that were evaluated on histopathology. ⁸⁹Zr-pan PET/CT had higher AUC (0.82, 95% CI, 0.68–0.95) compared to ¹⁸F-FDG PET/CT (0.76, 95% CI, 0.59–0.93; Table 2). Based on the optimal SUV_max cutoff of 2.61 and 9.27 for ⁸⁹Zr-pan PET/CT and ¹⁸F-FDG PET/CT using Youden's Index, respectively, ⁸⁹Zr-pan PET/CT had 100% specificity compared to 81.2% (95% CI, 62.1–100%) for ¹⁸F-FDG PET/CT (Table 2). However, ⁸⁹Zr-pan PET/CT had a sensitivity of 59.1% (95% CI, 38.5–79.6%) compared to 77.3% (95% CI, 59.8–94.8%) for ¹⁸F-FDG PET/CT. This suggests that the imaging modalities may have improved accuracy when performed together in the preoperative staging [sensitivity 60.9% (95% CI, 40.9–80.8%), specificity 100% (95% CI, 100–100%)].

Evaluation of whole-body PET/CT scans further reinforced the high specificity of ⁸⁹Zr-pan PET/CT in preoperative detection and staging of HNSCC as shown in Fig. 4A–E. Patient 2 presented with bilateral cutaneous SCC of the scalp and regional cervical adenopathy and underwent ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT prior to surgery. On ¹⁸F-FDG PET/CT, the patient was noted to have numerous mediastinal LNs (Fig. 4A; green outline) and a left inguinal node (Fig. 4B; green outline), which were not visualized on ⁸⁹Zr-pan PET/CT. The mediastinal nodes (biopsy) and other sites (18-month follow-up) were negative for SCC. Another patient had high SUV_max signals in a right apical pulmonary lesion and around a shoulder prosthesis on ¹⁸F-FDG PET/CT, (Fig. 4C and D; green outline), neither of which were visualized on ⁸⁹Zr-pan PET/CT. In the bottom row, a patient had good clinical response with neoadjuvant immunochemotherapy (Fig. 4E). ¹⁸F-FDG PET/CT demonstrated moderate signal at the primary tumor site but resolution of the cervical LNs. However, ⁸⁹Zr-pan PET/CT performed just prior to surgery was negative for any SUV_max signal, and there was no histopathological evidence of cancer in any of the specimens.

ROC curves were evaluated using the SUV_max for tumor and incidental lesions for ¹⁸F-FDG PET/CT (Fig. 4F), ⁸⁹Zr-pan PET/CT (Fig. 4G), and the combination of both scans (Fig. 4H). Forty-eight lesions (primary tumor, LNs, incidental findings) with median SUV_max 8.52 (range 2.0–23.6) were visualized on ¹⁸F-FDG PET/CT and 34 lesions with median SUV_max 2.0 (range 0.9–10.5) were identified with ⁸⁹Zr-pan PET/CT (detailed flowchart of identified lesions available in Supplementary Fig. S1). The combined ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT scans detected cancer lesions with a specificity of 96.3% (95% CI, 89.2%–100%) and AUC of 0.91 (95% CI, 0.84–0.99) compared with ¹⁸F-FDG PET/CT alone with specificity 74.1% (95% CI, 57.5%–90.6%) and AUC 0.79 (95% CI, 0.65–0.93).

There was no statistically significant difference in SUV_max between the LNs with and without cancer on pathologic assessment for ¹⁸F-FDG PET/CT (P = 0.064, one-way ANOVA). Among the 25 LNs identified as having high SUV_max signal on either PET/CT scan, 4 LNs were noted to have only ¹⁸F-FDG uptake, 20 LNs had both ¹⁸F-FDG and ⁸⁹Zr uptake, and 1 LN had only ⁸⁹Zr uptake, which was noted to be positive for disease on review of pathology (Fig. 5A). There was a significant difference in SUV_max between LNs with and without cancer on the ⁸⁹Zr-pan PET/CT scan (P = 0.029; one-way ANOVA), which correlated with a significant difference in MFI and SBR between LNs with and without cancer on fluorescence imaging (P < 0.001; Kruskal–Wallis; Fig. 5B). The specificity was higher for ⁸⁹Zr-pan PET/CT (86.7%; 95% CI, 69.5% – 100%) compared to ¹⁸F-FDG PET/CT (66.7%; 95% CI, 42.8% – 90.5%) (Fig. 5C). The combined ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT scans detected lymph node lesions with a specificity of 80% (95% CI, 59.8%–100%) and AUC of 0.81 (95% CI, 63.9–98.7) compared to ¹⁸F-FDG PET/CT alone [see above for specificity; AUC 0.73 (95% CI, 50.6–96.1)].

The liver received the highest organ dose, with a mean of 1.03 ± 0.38 mGy/MBq, and the maximum observed value was 1.58 mGy/MBq. The small intestine and the upper large intestine received a mean of 0.54 ± 0.15 and 0.53 ± 0.16 mGy/MBq respectively (maximum: 0.70 and 0.68 mGy/MBq) and kidneys and spleen received a mean dose of 0.40 ± 0.04 and 0.29 ± 0.07 mGy/MBq, respectively (maximum dose: 0.44 and 0.38 mGy/MBq). Absorbed doses in the remainder of the body were 0.14 ± 0.03 mGy/MBq (Supplementary Table S2).

This study demonstrates that ⁸⁹Zr-pan is a safe and specific radiotracer that can be used to improve accuracy in preoperative detection and staging of HNSCC, particularly in combination with ¹⁸F-FDG PET/CT. Although this was a phase 1 pilot study, we show limited/no adverse effects associated with the study drug and a high specificity and positive predictive value of ⁸⁹Zr-pan PET/CT to detect HNSCC lesions in the head and neck region, as well as in the whole body. The data on the combined use of ¹⁸F-FDG PET/CT and ⁸⁹Zr-pan PET/CT showed high specificity compared with ¹⁸F-FDG PET/CT alone, suggesting an opportunity to improve diagnostic accuracy and reduce the use of invasive diagnostic procedures. Moreover, this method could obviate invasive procedures necessary to confirm histology in lesions detected by ¹⁸F-FDG PET/CT coupled with the obvious advantage of decreasing risk to the patient and improve cost-effectiveness.

Pan800 has previously been studied in HNSCC and has been shown to have high sensitivity and specificity for detecting HNSCC in diverse applications, including intraoperative detection of the sentinel margin, assessment of the wound bed for residual tumor, and distinguishing between benign and metastatic LNs (18–20, 26). We independently evaluated two HNSCC-specific tracers, ⁸⁹Zr-pan and pan800, in our study. We found several similarities between the IRDye800- and ⁸⁹Zr- labeled panitumumab: there was generally higher signal in the primary tumor compared to the metastatic LNs (Fig. 3B), and there was a high degree of correlation between each imaging modality. This correlation between pan800 and ⁸⁹Zr-pan allows us to infer that ⁸⁹Zr-pan and pan800 have similar discriminating ability in distinguishing cancer vs. non-cancer. Using these combined preoperative and intraoperative modalities, a unique hybrid approach could be taken in which preoperative ⁸⁹Zr-pan PET/CT information could be validated in the operating room through imaging of pan800, thereby allowing the surgeon to confirm target lesion(s) seen on preoperative imaging.

We relied on the quantifiable PET signal, the SUV_max to determine the sensitivity and specificity of ⁸⁹Zr-pan. Previous studies evaluating ¹⁸F-FDG PET/CT used a SUV_max cut-off of 3.5 for detecting nodal metastases from oral SCC with (sensitivity 67.9, specificity 94.6; ref. 27). For ⁸⁹Zr-pan PET/CT, this cut off was found to be 2.6 (sensitivity 59.1, specificity 100) from our pilot study. The optimal scan time and dose for ⁸⁹Zr-pan remains undefined. However, based on our current study, the SBR using muscle as background begins to decrease after Day 3, suggesting that the optimal scan time is prior to Day 3 (Supplementary Fig. S2B).

⁸⁹Zr has been used by multiple research groups to label antibodies given its favorable half-life (28–31). The higher radiation dose was accommodated with a 1 mCi the maximum administered activity, which is a limitation. However, there are no alternative radioisotopes currently for in vivo PET imaging of antibodies. In order to improve image quality given the limited administered activity, longer scanning at each bed position, and using modern PET scanners with increased axial Field of View and SiPM detectors could result in improved image quality. Compared to the study by Lindenberg and colleagues (24), we showed decreased absorbed dose in the liver and spleen (P = 0.06, Kruskal–Wallis test; Supplementary Table S2), most likely due to a loading dose of “cold” panitumumab (pan800) in our study. These findings are in line with previous studies, in which a loading dose was demonstrated to decrease radiation exposure to organs as well as increased SUV_max values resulting in improved visualization of target lesions (31–39). Despite the loading dose, the major limitation of ⁸⁹Zr-pan PET/CT is the relatively high SUV_max signal in the liver, a consequence of the liver acting as a sink due to its high EGFR expression in hepatocytes in adult patients (40), which in turn also limits the dose of ⁸⁹Zr-pan that can be given and the image quality that can be achieved. Moreover, this also limits the ability to detect liver metastases. Fortunately, in HNSCC the highest incidence of distant metastases occurs in the lungs (up to 78%), bones (up to 30%) and the liver (17%; refs. 41, 42); thus, ⁸⁹Zr-pan PET/CT has the opportunity to further lesions after detection of the initial lesions with the current standard of care imaging. However, ⁸⁹Zr-pan may have limited application in other solid tumor with frequent metastases to the liver, such as pancreatic cancer (43). In addition, the longer half-life (compared with ¹⁸F-FDG) of ⁸⁹Zr-pan requires systemic administration of the drug up to 3 days prior to imaging, which may result in logistical challenges for routine clinical use. Further studies with increased sample size to determine the optimal dose and imaging window are needed to fully elucidate the potential value added of ⁸⁹Zr-pan PET/CT as a tumor imaging adjunct to the current standard of care of ¹⁸F-FDG PET/CT.

A. Iagaru reports grants from GE Healthcare, personal fees from ITM, Clarity, and Telix, and grants and personal fees from Novartis outside the submitted work. E.L. Rosenthal reports grants from NIH during the conduct of the study. No disclosures were reported by the other authors.

Y.J. Lee: Data curation, software, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. N.S. van den Berg: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, project administration, writing–review and editing. H. Duan: Data curation, formal analysis, validation, investigation, visualization, writing–review and editing. E. Azevedo: Validation, investigation, visualization, project administration. V. Ferri: Software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Hom: Data curation, validation, investigation, visualization, project administration, writing–review and editing. R.C. Raymundo: Investigation, methodology, project administration, writing–review and editing. A. Valencia: Data curation, validation, investigation, methodology, project administration, writing–review and editing. J. Castillo: Validation, investigation, methodology, project administration, writing–review and editing. B. Shen: Resources, supervision, validation, investigation, visualization, project administration, writing–review and editing. Q. Zhou: Investigation, visualization, methodology, writing–review and editing. L. Freeman: Data curation, investigation, visualization, project administration, writing–review and editing. M. Koran: Investigation, methodology, writing–review and editing. M.J. Kaplan: Data curation, validation, investigation, visualization, project administration, writing–review and editing. A. Colevas: Investigation, methodology, project administration, writing–review and editing. F.M. Baik: Validation, investigation, project administration, writing–review and editing. F.T. Chin: Resources, supervision, validation, investigation, project administration, writing–review and editing. B.A. Martin: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. A. Iagaru: Resources, data curation, software, formal analysis, supervision, investigation, visualization, methodology, project administration, writing–review and editing. E.L. Rosenthal: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing.

We thank Yifei Ma (Stanford University) for providing statistical consultation and Dr. Jared Grice (Vanderbilt University) for providing his expertise in medical physics. This work was supported in part by the Stanford Comprehensive Cancer Center (ELR), the Netherlands Organization for Scientific Research (NSVDB; Rubicon; 019.171LW.022), the National Institutes of Health and the National Cancer Institute (ELR; R01CA190306) and the National Institute on Deafness and Other Communication Disorders (YL; T32DC015209).

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.