HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Zhao, B. Articles by Kris, M. G. Search for Related Content PubMed PubMed Citation Articles by Zhao, B. Articles by Kris, M. G.

Lung Cancer: Computerized Quantification of Tumor Response—Initial Results¹

¹ From the Departments of Medical Physics and Radiology (B.Z.), Radiology (L.H.S., M.S.G.), Epidemiology and Biostatistics (C.S.M.), and Medicine (N.A.R., M.G.K.), Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021. Received November 18, 2005; revision requested January 5, 2006; revision received January 26; accepted February 8; final version accepted March 28. Supported in part by grants from William H. Goodwin and Alice Goodwin, the Commonwealth Foundation for Cancer Research, the Experimental Therapeutics Center of Memorial Sloan-Kettering Cancer Center, and National Cancer Institute (R21 CA113653-01 and PO1 CA005826-39). Address correspondence to B.Z. (e-mail: zhaob{at}mskcc.org).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 
Purpose: To prospectively quantify tumor response or progression in patients with lung cancer by using thin-section computed tomography (CT) and a semiautomated algorithm to calculate tumor volume and other parameter values.

Materials and Methods: This HIPAA-compliant study was institutional review board approved; informed patient consent was waived. CT scans of 15 measurable non–small cell lung cancers (in five men and 10 women; mean age, 64 years; range, 38–78 years) before and after gefitinib treatment were analyzed. A semiautomated three-dimensional lung cancer segmentation algorithm was developed and applied to each tumor at baseline and follow-up. The computer calculated the greatest diameter (unidimensional measurement), the product of the greatest diameter and the greatest perpendicular diameter (bidimensional measurement), and the volume of each tumor. Exact McNemar tests were used to analyze differences in the percentage change calculated with different measurement techniques.

Results: The computer accurately segmented 14 of the 15 tumors. One paramediastinal tumor required manual separation from the mediastinum. Eleven (73%) of the 15 patients had an absolute change in tumor volume of at least 20%, compared with one (7%) and four (27%) patients who had similar changes in unscaled unidimensional (P < .01) and bidimensional (P = .04) tumor measurements, respectively. Seven (47%) patients had an absolute change in tumor volume of at least 30%. In contrast, at unscaled analysis, no patients at unidimensional measurement (P = .02) and two (13%) patients at bidimensional measurement (P = .06) had a change of at least 30%.

Conclusion: Compared with the unidimensional and bidimensional techniques, semiautomated tumor segmentation enabled the identification of a larger number of patients with absolute changes in tumor volume of at least 20% and 30%.

© RSNA, 2006

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 
Evaluation of tumor response is critical for determining whether a particular treatment is effective in a patient or whether an experimental agent is effective against a specific tumor type. The standard way to assess the response of solid tumors to chemotherapy is to perform imaging—principally computed tomography (CT)—with use of unidimensional Response Evaluation Criteria in Solid Tumors (RECIST) or bidimensional World Health Organization (WHO) criteria (1–3) to measure the tumor's size. Neither the RECIST nor the WHO criteria include volume measurement, partly because of technical restrictions such as the anisotropic characteristics of diagnostic imaging and partly because of limitations of the available measurement methods (eg, hand-held calipers with film or electronic calipers with display monitors).

Current measurement techniques are limited by the inability to detect size changes in the axial direction, poor reproducibility due to complex disease morphology, and inter- and intraobserver variability (4–8). With the advent of thin-section CT, it is now possible to obtain image data sets with a spatial resolution adequate to measure tumor volumes. However, the radiologist faces challenges not only in producing accurate and reproducible measurements but also in performing measurements in a large number of tumors and carrying out efficient follow-up of the lesions to assess response to therapy.

Although isotropic imaging modalities are becoming more available in clinical settings and tumor volume measurements may enable more objective, accurate, and rapid assessment of therapy response than do traditional unidimensional and bidimensional techniques, further validation studies with large patient data sets of different cancer types are necessary. The development of computerized methods, both automated and semiautomated, for reliable quantification of tumor volume is essential to the successful validation of volumetric measurement techniques used to assess therapy responses and obtain routine diagnoses. Thus, the purpose of our study was to prospectively quantify tumor response or progression in patients with lung cancer by using thin-section CT and a semiautomated algorithm to calculate tumor volume and other parameter values.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 
Patient Data Collection
This Health Insurance Portability and Accountability Act–compliant study was approved by our institutional review board; informed patient consent was waived. At the time of this writing, 24 patients were enrolled in a Health Insurance Portability and Accountability Act–compliant, institutional review board–approved specific therapeutic clinical trial involving gefitinib (Iressa; Astra Zeneca Pharmaceuticals, Wilmington, Del). Patient informed consent had been obtained for this trial. Twenty-one of the 24 patients underwent biopsy; three patients had withdrawn their consent before biopsy. Fifteen of these 21 patients received a diagnosis of primary lung cancer and were considered to be eligible for the clinical therapeutic trial—that is, they had operable and resectable stage I or II non–small cell lung cancer, a smoking history of less than or equal to 10 pack-years, and/or bronchioalveolar carcinoma. These 15 patients (five men, 10 women; mean age, 64 years; range, 38–78 years) underwent gefitinib therapy for at least 21 days and were included in this prospective study. The mean time between the baseline and follow-up examinations was 26.4 days (range, 21–35 days). Thin-section CT was performed before and after gefitinib treatment for tumor response assessment.

CT Imaging
Multidetector CT was performed with a LightSpeed 16 scanner (GE Medical Systems, Milwaukee, Wis); the parameters used are given in Table 1. The images were obtained without intravenous contrast material injection during a breath hold of 7 seconds. The entire thorax, from the lower neck and including the adrenal glands in their entirety, was imaged. The acquired thin-section images were downloaded directly from the CT acquisition workstation to our research picture archiving and communication system server by way of the hospital's fast networking system, where the Digital Imaging and Communications in Medicine images were stored with the patient identification information removed (9,10). The images were then transferred to an Ultra 10 workstation (Sun Microsystems, Santa Clara, Calif) for segmentation.

The three-dimensional algorithm was modified to adjust itself for segmentation of larger metastases in the lungs and used in two studies related to oncologic therapeutic trials (13,14). For our current study, this three-dimensional algorithm was modified for better measurement of the primary lung cancers seen on multiple thin-section CT images. Instead of using the shape constraint to automatically determine the size of the morphologic filter used to detach adjacent blood vessels, as proposed with the earlier versions of the algorithm, we predefined two sizes of the morphologic filter with the aim of removing surrounding blood vessels up to 5 mm in diameter. To maximally suppress the inconsistency in tumor segmentation that might occur on baseline and follow-up CT images owing to adjacent tissues with an attenuation similar to that of the tumor, tumor pairs were processed simultaneously. For the pleural lesions, the strategy proposed by Zhao et al (15) was adopted: The lungs were extracted by using a thresholding algorithm followed by a morphologic closing operator to repair the distortions on the lung pleural surface caused by the lesions. The lesion segmentation then started in the extracted lungs.

If a large lesion attached to the mediastinum or to the combined right diaphragm and liver had an attenuation similar to the attenuations of these structures, the boundaries lying between the lesion and either the mediastinum or the liver needed manual separation. Thus, a free-hand region of interest on each lesion section would be generated before the segmentation algorithm was applied. Last, nodules with ground-glass opacity were segmented by using a fixed threshold followed by the morphologic opening operation.

One author (B.Z.) used this algorithm to segment all 15 tumors depicted on the CT images obtained before and after gefitinib treatment. After segmentation, the greatest diameter, greatest perpendicular diameter, and volume of each tumor on both images were automatically calculated by the computer (14).

All segmentation results, including the segmented lesions, greatest diameters, and greatest perpendicular diameters, were visually inspected by a board-certified radiologist (L.H.S.) who has 14 years of experience in CT image interpretation. The evaluations were performed by overlapping the contours of each segmented lesion on the original transverse images, which were then viewed carefully in cine mode. To determine the consistency of the three-dimensional algorithm, the segmentation results of each lesion on the baseline and follow-up CT images were displayed side by side. In addition, the two greatest axial diameters of each lesion were placed on the segmented lesion for the assessment. No case required manual modification because of a suboptimal segmentation result.

Statistical Analyses
The unidimensional (RECIST), bidimensional (WHO), and three-dimensional volumetric tumor measurements were compared in two ways. The first way was a direct comparison of the percentage changes in each measurement; we called this the unadjusted/unscaled method. The second way was a comparison of the measurements after conversion of the unidimensional and bidimensional measurements to an equivalent volume by using the standard methods that James et al described with illustrations in the description of the RECIST (16); we called this the adjusted/scaled method. For example, a 30% decrease in the unidimensional measurement correlates geometrically to a 50% decrease in the bidimensional measurement and a 65% decrease in the volumetric measurement. These geometric correlations are determined with the assumption that tumors are spherical and are the basis for the relationships between the unidimensional RECIST, the bidimensional WHO response criteria, and the suggested volumetric response guidelines outlined in the RECIST. Differences in measurements determined at the unadjusted/unscaled comparison reflect clear disparities. In contrast, differences in measurements determined after they are converted into an equivalent volume reflect mathematically pure changes that are due to the fact that tumors are not perfect spheres and do not change in a spherical manner.

Exact McNemar tests were used to test for differences in the percentage change calculated with different techniques. Because of the distribution of data that suggested possible outliers, Wilcoxon sign-rank tests were used to compare the continuous measurements. These analyses were conducted by using Stata 8.0 for Windows, 2003 (Stata, College Station, Tex).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 
Percentage changes in the unidimensional, bidimensional, and volumetric tumor measurements that occurred in the 15 patients during the interval between the baseline and follow-up CT examinations (Figs 1, 2) are shown in Figure 3; the specific measurements are given in Table 2.

View larger version (133K): [in this window] [in a new window] [Download PPT slide]   Figure 1: Semiautomated segmentation and quantification of a lung cancer on, A–D, baseline, and, E–H, follow-up CT images obtained in 64-year-old man (patient 3). A, Original transverse CT image shows manually selected nodule region of interest (enclosed in square). B, Automatically segmented tumor contour overlapped on each transverse region-of-interest image. C, Automatically determined greatest diameter and greatest perpendicular diameter displayed along with the segmented tumor on the section outlined (in black) in B. D, Three-dimensional depiction of the segmented tumor along the z-axis. E–H, Directly corresponding 5-week follow-up CT images of the same tumor segmented and quantified show the cancer to be almost unchanged at unidimensional measurement (C and G) (percentage change in size, –4.9%). However, obvious changes in the bidimensional (C and G) and volumetric (D and H) measurements are seen, with percentage changes in size of –30.8% and –41.6%, respectively.

View larger version (162K): [in this window] [in a new window] [Download PPT slide]   Figure 2a: Imaging findings in 38-year-old woman (patient 6) with no changes in unidimensional or bidimensional measurements but an enlarged tumor along the z-axis. (a) Baseline transverse CT image shows tumor contour (outlined in white) and the greatest diameter and the greatest perpendicular diameter (crossed lines) determined by using the semiautomated segmentation algorithm. (b) Three-dimensional image of the tumor segmented by using the semiautomated algorithm. (c, d) On corresponding follow-up CT images obtained 4 weeks later, the tumor is seen from the same angle along the z-axis. Tumor growth is identified in d only.

View larger version (37K): [in this window] [in a new window] [Download PPT slide]   Figure 2b: Imaging findings in 38-year-old woman (patient 6) with no changes in unidimensional or bidimensional measurements but an enlarged tumor along the z-axis. (a) Baseline transverse CT image shows tumor contour (outlined in white) and the greatest diameter and the greatest perpendicular diameter (crossed lines) determined by using the semiautomated segmentation algorithm. (b) Three-dimensional image of the tumor segmented by using the semiautomated algorithm. (c, d) On corresponding follow-up CT images obtained 4 weeks later, the tumor is seen from the same angle along the z-axis. Tumor growth is identified in d only.

View larger version (167K): [in this window] [in a new window] [Download PPT slide]   Figure 2c: Imaging findings in 38-year-old woman (patient 6) with no changes in unidimensional or bidimensional measurements but an enlarged tumor along the z-axis. (a) Baseline transverse CT image shows tumor contour (outlined in white) and the greatest diameter and the greatest perpendicular diameter (crossed lines) determined by using the semiautomated segmentation algorithm. (b) Three-dimensional image of the tumor segmented by using the semiautomated algorithm. (c, d) On corresponding follow-up CT images obtained 4 weeks later, the tumor is seen from the same angle along the z-axis. Tumor growth is identified in d only.

View larger version (43K): [in this window] [in a new window] [Download PPT slide]   Figure 2d: Imaging findings in 38-year-old woman (patient 6) with no changes in unidimensional or bidimensional measurements but an enlarged tumor along the z-axis. (a) Baseline transverse CT image shows tumor contour (outlined in white) and the greatest diameter and the greatest perpendicular diameter (crossed lines) determined by using the semiautomated segmentation algorithm. (b) Three-dimensional image of the tumor segmented by using the semiautomated algorithm. (c, d) On corresponding follow-up CT images obtained 4 weeks later, the tumor is seen from the same angle along the z-axis. Tumor growth is identified in d only.

View larger version (15K): [in this window] [in a new window] [Download PPT slide]   Figure 3: Unscaled/unadjusted percentage changes in unidimensional, bidimensional, and volumetric measurements of the 15 lung cancers during a predefined time of 3–5 weeks.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 
While debate on the advantages and disadvantages of RECIST and WHO guidelines continues (16–18), volumetric imaging technologies are approaching maturity and becoming available in clinical settings. The successful development of automated or semiautomated three-dimensional computer algorithms is essential to the ultimate validation and use of volumetric techniques for monitoring responses to treatment.

A prerequisite for obtaining size measurements—especially volume measurements—at serial thin-section CT is the automated or semiautomated (and reproducible) defining of the tumor's boundaries by means of separating the tumor from the surrounding normal tissues. Owing to the diversity of abnormalities, metastatic disease sites, and surrounding normal anatomic structures, the segmentation of lesions is very challenging; it requires the use of a priori knowledge and the development of task-specific strategies. In this study, we remodeled and applied a segmentation algorithm to measure the sizes of lung cancers in a clinical setting. The modified algorithm is based on an existing algorithm for segmenting small lung nodules depicted on CT images (11,12). The way the algorithm identifies nodule edges was unchanged: The edge is defined as the closed curve surrounding the lesion where the sharpest change in attenuation occurs around the boundary. Therefore, the modified algorithm had the same ability to identify a lesion's edge in regions where there were no abutting vascular structures.

If a lesion is attached to a surrounding soft tissue (eg, blood vessel) with similar attenuation, the soft tissue needs to be removed by using a morphologic filter. Although there is a default morphologic filter size in the modified algorithm, a larger filter can be chosen if the segmentation result is considered suboptimal. Larger morphologic filters were chosen manually in our study, and we anticipate further improvement in the algorithm with use of full automation. A limitation of our study was the lack of an independent reference standard for the tumor volume measurements.

Another factor that can affect volume measurements is the CT scanning technique used. Computer methods may perform differently in the identification of tumor contours on multiple CT images acquired with different parameters such as section thickness, pitch, reconstruction algorithm, peak voltage, and amperage. We specifically designed an imaging protocol for the evaluation of tumor volumes with use of our modified segmentation method. Investigating the effects of CT parameters on the performance of computer methods in the estimation of tumor volumes was beyond the scope of this study.

In our study, the unidimensional, bidimensional, and volumetric measurements of 15 measurable stage I or II non–small cell lung cancers depicted on baseline and follow-up CT images obtained 3–5 weeks apart were estimated by the computer. We selected cutoff values of 20% and 30% on an absolute scale because the RECIST criteria suggest that these are the unidimensional measurement values that should be used to identify tumor progression and response, respectively.

The differences in measurements seen at unadjusted/unscaled analysis in our study were greater than the differences seen at adjusted/scaled analysis. With either comparison method, the greatest changes were seen in the volumetric measurements. When the unadjusted scale was used, the changes seen differed significantly among the three measurement types. When the adjusted scale was used, the changes did not differ significantly among the three measurement types. These findings were not unexpected, because the unadjusted scale maximizes the differences between measurement methods by not correcting for the geometric scales. The differences seen at unadjusted/unscaled analysis reflect absolute differences in the measurement methods. Geometrically, a given percentage of unidimensional change in diameter corresponds to a larger change in volume. However, our unadjusted/unscaled analysis results show that the assumption that tumors are spherical is not always correct and that even when the known geometric relationship is taken into account, the various measurement methods may not yield equivalent results. Tumors do not necessarily have a spherical shape or change in a spherical manner.

Given the small sample size of 15 patients, differences between the volumetric estimations and the unidimensional and bidimensional measurements that were not significant at adjusted/scaled analysis were probably due to low statistical power. Further analysis of larger data sets is needed. Another limitation of our study is the fact that we do not yet have long-term follow-up data to assess whether patient benefit—measured in terms of survival or time to progression—correlates better with one scale or the other. Our study results demonstrate that significantly greater changes in tumor size may be accurately measured at volumetric response assessment in the unadjusted scale and that a similar trend is seen in the adjusted scale.

Our initial study results suggest that with use of thin-section CT images and computer software, changes in tumor volume can be assessed as early as 3 weeks after initiation of gefitinib treatment, whereas a lower magnitude of changes in unidimensional and bidimensional measurements is seen during the same time period. Thus, volumetric changes in tumor size may have the potential to be an earlier or better marker of regression or progression. Such a marker, by enabling a lack of tumor response to be recognized earlier or more reliably than it is recognized with conventional unidimensional and bidimensional measurements, could help to limit the amount of ineffective chemotherapy patients receive. As more therapeutic agents and approaches become available, it will become increasingly important to reevaluate existing methods and investigate technologies for determining lung cancer response at imaging.

	ADVANCE IN KNOWLEDGE

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 

• Semiautomated segmentation of lung cancer at baseline and at 3–5 weeks after treatment enabled the identification of a larger number of patients with absolute changes in tumor volume of at least 20% and 30%: 11 and seven patients, respectively, compared with one and no patients, respectively, identified with the unidimensional and bidimensional techniques.
  

	FOOTNOTES

 

Abbreviations: RECIST = Response Evaluation Criteria in Solid Tumors • WHO = World Health Organization

Authors stated no financial relationship to disclose.

Author contributions: Guarantors of integrity of entire study, B.Z., L.H.S.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, B.Z.; clinical studies, L.H.S., M.G., N.A.R., M.G.K.; experimental studies, B.Z., L.H.S.; statistical analysis, L.H.S., C.M.; and manuscript editing, B.Z., L.H.S., C.M., N.A.R., M.G.K.

	References

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCE IN KNOWLEDGE References

 

1. World Health Organization. WHO handbook for reporting results of cancer treatment. Offset publication no. 48. Geneva, Switzerland: World Health Organization, 1979.
2. Miller AB, Hogestraeten B, Staquet M, et al. Reporting results of cancer treatment. Cancer 1981;47:207–214.
3. Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate response to treatment in solid tumors. J Natl Cancer Inst 2000;92:205–216.
4. Thiesse P, Ollivier L, Di Stefano-Louineau D, et al. Response rate accuracy in oncology trial: reasons for interobserver variability. J Clin Oncol 1997;15:3507–3514.
5. Hopper KD, Kasales CJ, Van Slyke MA, et al. Analysis of interobserver and intraobserver variability in CT tumor measurements. AJR Am J Roentgenol 1996;167:851–854.
6. Lavin PT, Flowerdew G. Studies in variation associated with the measurement of solid tumors. Cancer 1980;46:1286–1290.
7. Belton AL, Saini S, Lieberman K, et al. Tumor size measurement in an oncology clinical trial: comparison between off-site and on-site measurements. Clin Radiol 2003;58:311–314.
8. Schwartz LH, Ginsberg MS, DeCorato D, et al. Evaluation of tumor measurements in oncology: use of film-based and electronic techniques. J Clin Oncol 2000;18:2179–2184.
9. Kalaigian J, Bidaut L, Zhao B, et al. Design and implementation of a flexible PACS interface for clinical research applications [abstr]. Radiology 2002;225(P):768.
10. Kalaigian JP, Schwartz LH, Kijewski PK, et al. Design and implementation of a targeted PACS infrastructure for clinical research [abstr]. In: Radiological Society of North America scientific assembly and annual meeting program. Oak Brook, Ill: Radiological Society of North America, 2004; 826.
11. Zhao B, Yankelevitz DF, Reeves AP, Henschke CI. Two-dimensional segmentation of pulmonary nodules on helical CT images. Med Phys 1999;26:889–895.
12. Zhao B, Reeves AP, Yankelevitz D, Henschke CI. Three-dimensional multi-criterion automatic segmentation of pulmonary nodules of helical CT images. Opt Eng 1999;38:1340–1347.
13. Zhao B, Schwartz LH, Lefkowitz RA, Wang L. Measuring tumor burden: comparison of automatic and manual techniques. In: Sonka M, Fitzpatrick J, eds. Proceedings of SPIE: medical imaging 2004—image processing. Vol 5370. Bellingham, Wash: International Society for Optical Engineering, 2004; 1695–1700.
14. Zhao B, Schwartz LH, Moskowitz CS, et al. Pulmonary metastases: effect of CT section thickness on measurement—initial experience. Radiology 2005;234:934–939.
15. Zhao B, Gamsu G, Ginsberg MS, Jiang L, Schwartz LH. Automated detection of small lung nodules on CT utilizing a local density maximum algorithm. J Appl Clin Med Phys 2003;4:248–260.
16. James K, Eisenhauer E, Christian M, et al. Measuring response in solid tumors: unidimensional versus bidimensional measurement. J Natl Cancer Inst 1999;91(6):523–528.
17. Husband JE, Schwartz LH, Spencer J, et al. Evaluation of the response to treatment of solid tumors: a consensus statement of the International Cancer Imaging Society. Br J Cancer 2004;90:2256–2260.
18. Mazumdar M, Smith A, Schwartz LH. A statistical simulation study finds discordance between WHO criteria and RECIST guideline. J Clin Epidemiol 2004;57:358–365.

This article has been cited by other articles:

				B. Zhao, L. P. James, C. S. Moskowitz, P. Guo, M. S. Ginsberg, R. A. Lefkowitz, Y. Qin, G. J. Riely, M. G. Kris, and L. H. Schwartz Evaluating Variability in Tumor Measurements from Same-day Repeat CT Scans of Patients with Non-Small Cell Lung Cancer Radiology, July 1, 2009; 252(1): 263 - 272. [Abstract] [Full Text] [PDF]

				M. A. Gavrielides, L. M. Kinnard, K. J. Myers, and N. Petrick Noncalcified Lung Nodules: Volumetric Assessment with Thoracic CT Radiology, April 1, 2009; 251(1): 26 - 37. [Abstract] [Full Text] [PDF]

				B. Zhao, L. H. Schwartz, and S. M. Larson Imaging Surrogates of Tumor Response to Therapy: Anatomic and Functional Biomarkers J. Nucl. Med., February 1, 2009; 50(2): 239 - 249. [Abstract] [Full Text] [PDF]

				W. M. Stadler Tumor Burden Endpoints and Phase II Clinical Trial Design ASCO Educational Book, January 1, 2008; 2008(1): 89 - 93. [Abstract] [Full Text] [PDF]

				G. J. Riely, M. G. Kris, B. Zhao, T. Akhurst, D. T. Milton, E. Moore, L. Tyson, W. Pao, N. A. Rizvi, L. H. Schwartz, et al. Prospective Assessment of Discontinuation and Reinitiation of Erlotinib or Gefitinib in Patients with Acquired Resistance to Erlotinib or Gefitinib Followed by the Addition of Everolimus Clin. Cancer Res., September 1, 2007; 13(17): 5150 - 5155. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Zhao, B. Articles by Kris, M. G. Search for Related Content PubMed PubMed Citation Articles by Zhao, B. Articles by Kris, M. G.