Diagnosis of Pseudoprogression Following Lomustine–Temozolomide Chemoradiation in Newly Diagnosed Glioblastoma Patients Using FET-PET

Glioblastoma is a devastating disease with median overall survival times in unselected study populations of around 15–20 months (1, 2). Importantly, the subgroup of patients with glioblastoma with a methylated O⁶-methylguanine-DNA methyltransferase (MGMT) promoter treated with temozolomide chemoradiation has a better prognosis with median survival times beyond 20 months (3). The prolonged survival is mainly due to the susceptibility of the MGMT promotor methylated glioblastoma cells to alkylating chemotherapy (4). The prospective CeTeG/NOA-09 phase-III trial recently demonstrated that the median overall survival can be further prolonged to approximately 48 months by lomustine–temozolomide chemoradiation as first-line therapy in patients with glioblastoma with methylated MGMT promoter (5).

However, chemoradiation with intensified alkylating chemotherapy using lomustine and temozolomide applied according to the CeTeG/NOA-09 trial protocol may cause prolonged and/or delayed pseudoprogression (6). For example, Stuplich and colleagues described that this phenomenon occurs substantially later than 12 weeks after radiotherapy (onset, 17–50 weeks; ref. 6). In contrast, pseudoprogression following standard chemoradiation with temozolomide alone occurs earlier, typically within the first 12 weeks after radiotherapy completion (7–9). Therefore, this time-dependent definition has been incorporated into the criteria defined by the Response Assessment in Neuro-Oncology Working Group (RANO criteria; ref. 7).

Furthermore, an earlier study reported an increased pseudoprogression rate in patients with glioma with methylated MGMT promoter (9). Most probably due to standardizing the assessment of response, tumor progression, and pseudoprogression using strict algorithms in subsequent phase III clinical trials (5, 10), this association could no longer be observed. On the other hand, one of the most extensive studies with more than 250 glioblastoma patients treated with temozolomide chemoradiation reported a 3.5-fold greater probability to develop pseudoprogression if a MGMT promoter methylation is present (11). Thus, the effect of an MGMT promoter methylation on the development of pseudoprogression cannot be neglected.

Neurooncologic centers, especially in Europe, are increasingly offering lomustine–temozolomide chemoradiation due to its survival-prolonging effects, which may also provoke a higher pseudoprogression rate. Importantly, this can lead to an erroneously premature termination of lomustine–temozolomide chemotherapy with a potential negative prognostic impact and the misinterpretation of response using conventional MRI. Such an incorrect diagnosis may also lead to an inadequate patient inclusion in clinical trials, eventually resulting in misleading data about the applied treatment for (falsely assumed) progressive disease, thereby overestimating efficacy.

Contrast-enhanced conventional MRI is the cornerstone of brain tumor imaging, but this technique suffers from low specificity despite excellent spatial resolution (12–14). The differentiation of treatment-related changes from actual tumor progression represents a major problem not only after chemoradiation with alkylating agents but also with other treatment options currently used in brain tumor patients (e.g., antiangiogenic therapy, checkpoint inhibitor immunotherapy, or targeted therapy, especially in combination with radiotherapy; refs. 15–17).

To overcome the limitations of conventional MRI, diagnostic tools with higher accuracy are necessary. In recent years, it has repeatedly been shown that in patients with glioma, PET using the radiolabeled amino acid O-(2-[¹⁸F]fluoroethyl)-L-tyrosine (FET) offers high diagnostic accuracy for differentiating pseudoprogression from actual tumor progression (18–23). In addition, the high clinical value of amino acid PET for identifying treatment-related changes in patients with primary and secondary brain tumors has been highlighted by the RANO Working Group (24, 25).

FET is not significantly incorporated into any metabolic pathway and has no relevant participation in protein synthesis and is transported via specific amino acid transporters especially by the system of l-type amino acid transporters (LAT), particularly the subtypes LAT1 and LAT2 (26). An experimental study suggested that the overexpression of LAT1 strongly facilitates the influx of FET. On the other hand, FET turned out to be a poor efflux substrate of LAT1. Thus, this asymmetry may cause the intracellular entrapment of FET (27).

We evaluated the diagnostic performance of static and dynamic FET-PET parameters for differentiating pseudoprogression from actual tumor progression in patients with glioblastoma with methylated MGMT promoter treated off-study with lomustine–temozolomide chemoradiation according to the CeTeG/NOA-09 protocol.

From 2018 to 2020, we retrospectively identified patients who (i) were treated off-study according to the CeTeG/NOA-09 protocol (5), (ii) had equivocal or progressive MRI findings after radiotherapy inside the radiation field (i.e., within the 80% isodose), and (iii) underwent additional FET-PET imaging for diagnostic evaluation (number of scans, 1–3).

According to the CeTeG/NOA-09 protocol (5), the patients underwent external fractionated radiotherapy (60 Gy, 30 fractions over 6 weeks) after resection or biopsy. In addition to radiotherapy, the patients received up to six cycles of lomustine (100 mg/m² on day 1) plus temozolomide (100–200 mg/m² per day on days 2–6 of the 6-week course). The first cycle of lomustine–temozolomide chemotherapy started in the first week of radiotherapy.

The local ethics committees approved the retrospective analysis of neuroimaging data. There was no conflict with the Declaration of Helsinki. Before PET imaging, all patients had given written informed consent for the PET investigation and the use of the data for scientific purposes.

If available, neuropathologic analysis after biopsy or resection at the time of progression was used to diagnose pseudoprogression or tumor progression. In those cases, the presence of viable tumor tissue confirmed tumor progression. On the other hand, pseudoprogression was considered if histology predominantly showed features typically associated with treatment effects such as bland necrosis, fibrosis, gliosis, edema, demyelination, and vascular hyalinization (28).

In cases without neuropathologic confirmation, pseudoprogression was assumed if after an initial MRI with equivocal or progressive lesion(s) according to RANO criteria (7): (i) MRI findings remained stable or subsequently improved during follow-up examinations performed every 8–12 weeks, (ii) the clinical status remained stable or improved again after initial deterioration during follow-up examinations performed every 8–12 weeks, and (iii) the treatment was not changed during the following 6 months. This definition was modified from Young and colleagues (29). Tumor progression was diagnosed if MRI changes, or clinical deterioration prompted a treatment change.

Following the International Standardized Brain Tumor Imaging Protocol (BTIP; ref. 30), MR imaging was performed using a 1.5 T or 3.0 T scanner before and after administration of a gadolinium-based contrast agent (0.1 mmol/kg body weight). The sequence protocol comprised at least 3D isovoxel T1-weighted, 2D T2-weighted, and 2D fluid-attenuated inversion recovery-weighted (FLAIR) sequences.

As described previously, the amino acid FET was produced via nucleophilic ¹⁸F-fluorination with a radiochemical purity of greater than 98%, specific radioactivity greater than 200 GBq/μmol, and a radiochemical yield of about 60% (19, 31). According to international guidelines for brain tumor imaging using radiolabeled amino acid analogues (32), all patients fasted for at least 4 hours before the PET measurements. All patients underwent a dynamic PET scan from 0 to 50 minutes postinjection of 3 MBq of FET per kg of body weight. PET imaging was performed either on an ECAT Exact HR+ PET scanner (n = 14 patients) in three-dimensional mode (Siemens; axial field-of-view, 15.5 cm; spatial resolution, 6 mm) or simultaneously with 3T MR imaging using a BrainPET insert (n = 9 patients; Siemens). The BrainPET is a compact cylinder that fits in the Magnetom Trio MR scanner (axial field of view, 19.2 cm; spatial resolution, 3 mm; ref. 33). Iterative reconstruction parameters were 16 subsets, 6 iterations using the OSEM algorithm for the ECAT HR+ PET scanner, and 2 subsets, 32 iterations using the OP-OSEM algorithm for the BrainPET. Data were corrected for random, scattered coincidences, dead time, and motion for both systems. Attenuation correction for the ECAT HR+ PET data was based on a transmission scan, and for the BrainPET data on a template-based approach (33). The reconstructed dynamic dataset consisted of 16 time frames (5 × 1 minutes; 5 × 3 minutes; 6 × 5 minutes) for both scanners.

To optimize comparability of the results related to the influence of the two different PET scanners, reconstruction parameters, and postprocessing steps, a 2.5-mm 3D Gaussian filter was applied to the BrainPET data before further processing as described previously (19). In phantom experiments using spheres of different sizes to simulate lesions, this filter kernel demonstrated the best comparability between PET data obtained from the ECAT HR+ PET and the BrainPET scanner (34).

Static FET-PET data analysis was based on summed PET images over 20–40 minutes postinjection. The tumor area on FET-PET scans was determined by a two-dimensional autocontouring process using a tumor-to-brain ratio (TBR) of 1.6 or more as described previously (19, 35, 36). Maximum and mean tumor-to-brain ratios (TBR_max, TBR_mean) were calculated by dividing the maximum and mean standardized uptake value (SUV) of the tumor regions of interest by the mean SUV of healthy brain tissue.

As described previously (19, 37), time–activity curves (TAC) of the mean FET uptake in the tumor were generated by applying a spherical volume-of-interest (VOI) with a volume of 2 mL centered on the voxel with the maximum tumor uptake to the entire dynamic dataset. A reference TAC was generated by placing a reference ROI in the unaffected brain tissue as reported (37). For TAC evaluation, the time-to-peak (TTP; time in minutes from the beginning of the dynamic acquisition up to the maximum SUV of the lesion) was determined. In cases with steadily increasing FET uptake without identifiable peak, we defined the end of the dynamic PET acquisition as TTP. Furthermore, the TAC slope in the late phase of FET uptake was assessed by fitting a linear regression line to the late phase of the curve (20–50 minutes postinjection). The slope was expressed as the change of the SUV per hour. This procedure allows for a more objective evaluation of kinetic data than an assignment of TACs to earlier reported patterns of FET uptake during dynamic acquisition (37).

In addition to evaluating a single FET-PET scan corresponding to the suspicious MRI, the diagnostic value of relative TBR_mean changes was also evaluated in patients with more than one FET-PET scan. A previous study has suggested that a decrease of the TBR_mean of 10% or more is of value to diagnose pseudoprogression in patients with glioblastoma treated with standard temozolomide chemoradiation (38). Accordingly, in patients with a reduction of TBR_mean of 10% or more than the reference FET-PET scan without clinical deterioration, pseudoprogression was considered. Analogously, tumor progression was considered in case of an increase of TBR_mean > 10% compared with the reference FET-PET scan.

As described previously (39), all tumors were neuropathologically classified according to the WHO Classification of Tumors of the Central Nervous System of 2016 (40). For molecular biomarker analysis, tumor DNA was extracted from formalin-fixed and paraffin-embedded tissue samples with a histologically estimated tumor cell content of 80% or more. For assessment of the IDH mutation status, the presence of an IDH1-R132H mutation was evaluated by IHC using a mutation-specific antibody in a standard IHC staining procedure as reported before (39, 41, 42). When immunostaining for IDH1-R132H remained negative, the mutational hotspots at codon 132 of IDH1 and codon 172 of IDH2 were directly sequenced, as reported previously (42, 43). The 1p/19q codeletion status was determined by PCR-based microsatellite analysis as reported (44). The O⁶-methylguanine-DNA-methyltransferase (MGMT) promoter methylation status was assessed by methylation-specific PCR as described elsewhere (43).

Descriptive statistics are provided as mean and SD or median and range. For intergroup comparison, the Student t test for independent samples was used when variables were normally distributed, and the Mann–Whitney U test if variables were not normally distributed. ROC curve analyses were performed to define the decision cut-off values for static and dynamic FET-PET parameters using the neuropathologic confirmation or clinicoradiologic course as reference. The decision cut-off was considered optimal when the product of paired values for sensitivity and specificity reached its maximum. Moreover, the area under the ROC curve (AUC), its SE, and level of significance were determined to measure the test's diagnostic quality. The Fisher exact test for 2 × 2 contingency tables was used to evaluate the diagnostic performance of FET-PET parameters. P values of 0.05 or less were considered significant. Statistical analyses were performed using GraphPad Prism (Release 8.4.3, GraphPad Software Inc.).

According to our search criteria, we identified 23 adult patients (mean age, 58 ± 9 years; age range, 38–71 years; 10 females) with 32 FET-PET scans and neuropathologically confirmed IDH-wild-type glioblastoma and methylated MGMT promoter. In addition, 7 patients were identified who underwent dynamic FET-PET imaging before initiating lomustine–temozolomide chemoradiation. Another patient underwent additional dynamic FET-PET imaging before the initiation of lomustine–temozolomide chemoradiation and 7 weeks after the suspicious MRI.

After resection (n = 12 complete resections and n = 7 partial resections) or stereotactic biopsy (n = 4), all patients underwent external fractionated radiotherapy (60 Gy) plus lomustine–temozolomide chemotherapy according to the CeTeG/NOA-09 trial (5).

After radiotherapy completion, all patients exhibited equivocal MRI findings suspicious for tumor progression. Fifteen patients (65%) had a new contrast-enhancing lesion (n = 9) or an enlargement of a preexisting contrast-enhancing lesion of at least 25% (n = 6) according to the RANO criteria (7). Four patients had an enlargement of a preexisting contrast-enhancing lesion of less than 25%, and 1 patient had a new contrast-enhancing lesion with a maximum diameter smaller than 10 mm. Three patients had a signal increase of the perifocal hyperintensity on the T2- or FLAIR-weighted MRI of 15% or more (45). Patients' characteristics and MRI findings are listed in Table 1.

The mean time between completion of radiotherapy and the MRI suspicious for tumor progression was 14 ± 9 weeks (median time, 10 weeks; range, 5–34 weeks). All patients underwent dynamic FET-PET imaging to differentiate between pseudoprogression and actual tumor progression after a mean time of 10 ± 7 days (median time, 11 days; range, 0–26 days).

In 11 of 23 patients (48%), pseudoprogression was diagnosed within 5–25 weeks after radiotherapy completion (median time, 9 weeks). In 9 patients, a stable clinical status combined with stable or improved MRI findings during a follow-up of more than 6 months (median follow-up, 12 months; range, 6–31 months) without treatment change confirmed pseudoprogression. In 7 of these 9 patients, the median time between pseudoprogression onset and the beginning of the improvement of MRI findings was 34 weeks (range, 26–99 weeks). One patient example is presented in Fig. 1. In 2 patients, the neuropathologic tissue examination of the suspicious lesion revealed no viable tumor cells, and reactive tissue changes confirmed pseudoprogression. One patient example is presented in Fig. 2.

Tumor progression was confirmed in 12 of 23 patients (52%). In 8 patients, tumor progression was diagnosed clinicoradiologically (median follow-up, 2 months; range, 0–5 months). In the remaining 4 patients, the obtained neuropathologic samples yielded predominantly viable tumor cells. A summary of results is provided in Table 2.

The median number of applied lomustine–temozolomide chemotherapy cycles before the occurrence of equivocal or progressive findings on MRI was 3 (range, 2–6 cycles). In patients with pseudoprogression or actual tumor progression, there were no significant differences in terms of age (57 ± 9 vs. 59 ± 10 years; P = 0.631) or the extent of resection (complete resection rate in patients with pseudoprogression, 55%; complete resection rate in patients with actual tumor progression, 50%).

The diagnostic performance of conventional MRI based on RANO criteria (7) for the identification of pseudoprogression did not reach significance (accuracy, 58%; sensitivity, 30%; specificity, 79%; P = 0.665).

The static FET uptake parameters TBR_mean and TBR_max were significantly lower in patients with pseudoprogression compared with patients with actual tumor progression (TBR_mean, 1.9 ± 0.2 vs. 2.1 ± 0.2; P = 0.023, and TBR_max, 2.8 ± 0.6 vs. 3.2 ± 0.5; P = 0.045). The dynamic PET parameter TTP was significantly higher in patients with pseudoprogression than with actual tumor progression (36.6 ± 8.3 vs. 24.8 ± 9.4 minutes; P = 0.005). Regarding the dynamic parameter slope, group comparison did not reach significance (P > 0.05). Results are summarized in Table 2.

ROC analysis revealed that the optimal cut-off value of TBR_mean for the differentiation of pseudoprogression from actual tumor progression was 1.95 (sensitivity, 82%; specificity, 92%; accuracy 87%; AUC, 0.77 ± 0.12; P = 0.029). The corresponding TBR_max cut-off value was 2.85 (sensitivity, 64%; specificity, 92%; accuracy 78%; AUC, 0.75 ± 0.11; P = 0.046). Regarding the dynamic parameter TTP, ROC analysis revealed an optimal cut-off value of 35 minutes (sensitivity, 64%; specificity, 83%; accuracy 74%; AUC 0.82 ± 0.09; P = 0.010). The dynamic parameter slope yielded no statistically significant results.

Furthermore, the combination of TBR_mean with TTP increased both the specificity and positive predictive value to 100% for the detection of pseudoprogression (sensitivity, 55%; accuracy 78%; P = 0.005). An overview of the results is provided in Table 3.

Relative changes of TBRs calculated from additional FET-PET scans (n = 9) of 8 patients provided additional diagnostic information. In 5 of these 8 patients, relative FET-uptake changes (i.e., > 10% decrease or increase of TBRs) enabled the diagnosis of pseudoprogression or tumor progression in 3 and 2 patients, respectively. The integration of relative TBR_mean changes yielded an improved diagnostic accuracy of 91% for identifying pseudoprogression (sensitivity, 91%; specificity, 92%; P < 0.001). The relative change of TBR_max did not improve diagnostic accuracy.

The main finding of this study is that FET-PET–derived imaging parameters seem to be of significant clinical value for detecting pseudoprogression in newly diagnosed IDH-wild-type glioblastoma with methylated MGMT promoter treated with lomustine–temozolomide chemoradiation. Consequently, suspicious MRI findings during that treatment regimen should be critically evaluated.

As described previously, FET-PET–derived parameters show a high diagnostic accuracy for detecting treatment-related changes such as pseudoprogression or radiation necrosis following chemoradiation with standard alkylating agents (i.e., predominantly temozolomide; refs. 18–20, 22, 37). These results extend these findings, suggesting that both static and dynamic FET-PET parameters are also of clinical value for detecting pseudoprogression related to lomustine–temozolomide chemoradiation. Regarding static parameters, the herein reported threshold of 1.95 for TBR_mean and the resulting high diagnostic accuracy are following earlier studies (18, 19, 46).

Furthermore, and in line with previous studies (19, 37, 46), we observed that dynamic FET parameters also have an additional diagnostic value for pseudoprogression detection. As outlined above, it is assumed that the different FET uptake kinetics is caused by a differential asymmetry of influx and efflux of FET in progressive tumor and pseudoprogression. In particular, the combination of static and dynamic FET-PET parameters increased both the specificity and positive predictive value to 100%. Generally, conventional contrast-enhanced MRI provides high sensitivity, but its specificity for diagnosing treatment-related changes is limited (12). However, especially a high specificity is essential for the correct detection of pseudoprogression and can be improved by static and dynamic FET-PET parameter combinations.

Finally, in addition to FET-PET parameter combinations, our study also highlights the value of relative changes of metabolic activity using serial FET-PET imaging for diagnosing pseudoprogression. As observed in an earlier study (38), a reduction of TBR_mean of more than 10% was also associated with improved diagnostic accuracy for pseudoprogression.

For the management of patients with glioblastoma, another important issue is the onset of pseudoprogression following lomustine–temozolomide chemoradiation. In the majority of patients, we observed the onset of pseudoprogression within the first 12 weeks after radiotherapy completion. Notwithstanding, in our study, pseudoprogression was diagnosed in 2 patients beyond the 12-week time window, indicating that pseudoprogression may also occur delayed. This observation is in line with the results of a study by Stuplich and colleagues. In that study, late and prolonged pseudoprogression was observed in 3 of 8 patients treated with lomustine–temozolomide chemoradiation, and the pseudoprogression onset was between 19–50 weeks after radiotherapy completion (6). Furthermore, the colleagues reported a slow improvement of contrast-enhancing lesions within 41–97 weeks, indicating prolonged pseudoprogression. In a more recent study, in 2 patients with glioblastoma undergoing lomustine–temozolomide chemoradiation, late pseudoprogression occurred 18–24 weeks after radiotherapy completion (20). However, information on the improvement of contrast-enhancing lesions was not reported in all patients. In this study, the latest pseudoprogression onset was 25 weeks after radiotherapy, and the most extended time interval between pseudoprogression onset and improvement of MRI findings was 99 weeks.

Following standard chemoradiation with temozolomide in patients with glioblastoma, previous studies suggested that pseudoprogression occurs most frequently within the first 12 weeks after chemoradiation completion (9, 11, 47–49). In contrast, the rate of late pseudoprogression beyond the 12-week window following standard temozolomide-based chemoradiation is relatively low (< 5%) and observed in single cases only (47, 50–52). However, it is unclear why pseudoprogression was late or prolonged beyond the 12-week cut-off after radiotherapy. A possible explanation may be the long-lasting effects of lomustine (53). For example, Peyre and colleagues observed in 60% of 21 patients with glioma treated with lomustine, procarbazine, and vincristine that a prolonged response can be detected more than 3 years after treatment discontinuation (median period, 2.7 years; ref. 53). Accordingly, these long-lasting effects of nitrosoureas, such as lomustine, may also be responsible for late and prolonged pseudoprogression.

A few limitations of our study need to be discussed. Because of the retrospective character of this study, the results need to be confirmed prospectively. Another potential weakness is the relatively small number of patients. On the other hand, we included only patients with IDH-wild-type glioblastoma, and this homogenous group of patients was balanced in terms of resection extent and gender distribution. Note that a neuropathologic validation was not available for all lesions, and in most lesions, clinicoradiologic criteria had to be used for the definite diagnosis. Nevertheless, due to a poor clinical condition or refusal of the patient, another biopsy or surgery could not always be performed.

It should be noted that advanced MRI techniques such as perfusion- and diffusion-weighted MRI as well as proton MR spectroscopy are also of clinical value for the differentiation of tumor progression from pseudoprogression (13). In the recent years, methods from the field of artificial intelligence such as feature-based radiomics are increasingly used for the diagnosis of pseudoprogression. Available study results, mainly based on both conventional and advanced MRI radiomics, are promising (54). Moreover, initial studies evaluating feature-based radiomics computed from FET-PET and FET-PET evaluation methods based on machine learning suggested also an improvement of diagnostics (55, 56).

In summary, our study suggests that FET-PET is of significant clinical value for diagnosing late and prolonged pseudoprogression following lomustine–temozolomide chemoradiation according to the CeTeG/NOA-09 trial. An incorrect diagnosis of tumor progression (i.e., overlooking pseudoprogression) implies that an effective treatment may be erroneously terminated, with a potentially harmful influence on survival. The latter is of particular relevance because treatment options after lomustine–temozolomide chemoradiation are limited, as alkylating agents are usually exhausted. Therefore, additional FET-PET imaging should be considered in patients with glioblastoma treated with radiation and intensified alkylating chemotherapy in the case of equivocal MRI findings.

N. Schäfer reports personal fees from Bayer outside the submitted work. G.R. Fink reports personal fees from Bayer, Desitin, DGN, Ergo DKV, FomF, GSK, Medica Academy Messe Düsseldorf, Medicbrain Healthcare, Novartis, Pfizer, and Sportärztebund NRW outside the submitted work. U. Herrlinger reports personal fees from Novartis, Daichii-Sankyo, Noxxon, AbbVie, Bayer, Janssen, and Karyopharm, as well as personal fees from Medac outside the submitted work. No disclosures were reported by the other authors.

J.-M. Werner: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Weller: Data curation, investigation, writing–review and editing. G. Ceccon: Data curation, writing–review and editing. C. Schaub: Data curation, writing–review and editing. C. Tscherpel: Data curation, writing–review and editing. P. Lohmann: Formal analysis, investigation, writing–review and editing. E.K. Bauer: Data curation, writing–review and editing. N. Schäfer: Data curation, writing–review and editing. G. Stoffels: Data curation, investigation, writing–review and editing. C. Baues: Data curation, writing–review and editing. E. Celik: Data curation, writing–review and editing. S. Marnitz: Data curation, writing–review and editing. C. Kabbasch: Formal analysis, investigation, writing–review and editing. G.H. Gielen: Visualization, writing–review and editing. G.R. Fink: Supervision, funding acquisition, investigation, writing–review and editing. K.-J. Langen: Formal analysis, investigation, writing–review and editing. U. Herrlinger: Formal analysis, investigation, writing–review and editing. N. Galldiks: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The Cologne Clinician Scientist-Program (CCSP) of the Deutsche Forschungsgemeinschaft (DFG, FI773/15-1), Germany, supported this work.

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.