Suitability of Semi-Automated Tumor Response Assessment of Liver Metastases using a Dedicated Software Package

Zusammenfassung

Ziel: Ziel war es, den Nutzen einer Software für die semiautomatische Beurteilung des Tumoransprechens (tumor response assessment, TRA) von Lebermetastasen im Vergleich zur manuellen Methode auszuwerten. Material und Methoden: Insgesamt wurden 32 Patienten mit hepatisch metastasiertem kolorektalem Karzinom durch im Mittel 2,8 CT-Untersuchungen mit Kontrastmittel kontrolliert. Zwei Radiologen (O1, O 2) bestimmten den Längsdurchmesser (LD) von 96 Metastasen im Verlauf manuell sowie semiautomatisch durch die mit dem PACS verknüpfte Software (LMS-Liver, MEDIAN Technologies). LD und TRA („progressive”, „stable”, „partial remission”) wurden in Anlehnung an RECIST (Response Evaluation Criteria in Solid Tumors) ermittelt und hinsichtlich Variabilität zwischen den Methoden, Inter- und Intraobservervariabilität untersucht. Die benötigte Zeit für die Evaluation des TRA wurde zwischen den Methoden verglichen. Ergebnisse: Insgesamt korrelierten die LD-Messungen exzellent (r ≥ 0,96). Der Intraobserver- (semiautomatisch), Interobserver- (manuell) sowie Methodenvergleich (durch O 1) ergab bei mittleren Differenzen von 1,4 ± 2,6 mm, 1,9 ± 1,9 mm bzw. 2,1 ± 2,0 mm keine signifikante Variabilität. Im Interobserver- (semiautomatisch) sowie Methodenvergleich (durch O 2) reflektierten die LD-Differenzen von im Mittel 3,0 ± 3,0 mm bzw. 2,6 ± 2,0 mm eine signifikante Variabilität (p < 0,01). Manuelle und semiautomatische TRA stimmten zwischen den Beobachtern in 91,4 % und im Intraobserververgleich (semiautomatisch) in 84,5 % überein. Zwischen den Methoden stimmte das TRA in je 86,2 % überein. Die semiautomatische (2,7 min.) dauerte etwas länger als die manuelle Auswertung (2,3 min). Schlussfolgerung: Semiautomatische und manuelle Auswertung erlauben eine vergleichbar gute Einschätzung des Tumoransprechens von Lebermetastasen und erfordern einen ähnlichen Aufwand.

Abstract

Purpose: To evaluate the suitability of semi-automated compared to manual tumor response assessment (TRA) of liver metastases. Materials and Methods: In total, 32 patients with colorectal cancer and liver metastases were followed by an average of 2.8 contrast-enhanced CT scans. Two observers (O1, O 2) measured the longest diameter (LD) of 269 liver metastases manually and semi-automatically using software installed as thin-client on a PACS workstation (LMS-Liver, MEDIAN Technologies). LD and TRA (”progressive”, ”stable”, ”partial remission”) were performed according to RECIST (Response Evaluation Criteria in Solid Tumors) and analyzed for between-method, interobserver and intraobserver variability. The time needed for evaluation was compared for both methods. Results: All measurements correlated excellently (r ≥ 0.96). Intraobserver (semi-automated), interobserver (manual) and between-method differences (by O 1) in LD of 1.4 ± 2.6 mm, 1.9 ± 1.9 mm and 2.1 ± 2.0 mm, respectively, were not significant. Interobserver (semi-automated) and between-method (by O 2) differences in LD of 3.0 ± 3.0 mm and 2.6 ± 2.0 mm, respectively, reflected a significant variability (p < 0.01). The interobserver agreement in manual and semi-automated TRA was 91.4 %. The intraobserver agreement in semi-automated TRA was 84.5 %. Between both methods a TRA agreement of 86.2 % was obtained. Semi-automated evaluation (2.7 min) took slightly more time than manual evaluation (2.3 min). Conclusion: Semi-automated and manual evaluation of liver metastases yield comparable results in response assessments and require comparable effort.

Key words

CT spiral - liver metastases - tumor response - RECIST - software - semi-automated segmentation

References

• 1 Miller A B, Hoogstraten B, Staquet M. et al . Reporting results of cancer treatment.  Cancer. 1981;  47 207-214
• 2 Therasse P, Arbuck S G, Eisenhauer E A. et al . New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada.  J Natl Cancer Inst. 2000;  92 205-216
• 3 Eisenhauer E A, Therasse P, Bogaerts J. et al . New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1).  Eur J Cancer. 2009;  45 228-247
• 4 Marten K, Auer F, Schmidt S. et al . Inadequacy of manual measurements compared to automated CT volumetry in assessment of treatment response of pulmonary metastases using RECIST criteria.  Eur Radiol. 2006;  16 781-790
• 5 Erasmus J J, Gladish G W, Broemeling L. et al . Interobserver and intraobserver variability in measurement of non-small-cell carcinoma lung lesions: implications for assessment of tumor response.  J Clin Oncol. 2003;  21 2574-2582
• 6 Belton A L, Saini S, Liebermann K. et al . Tumour size measurement in an oncology clinical trial: comparison between off-site and on-site measurements.  Clin Radiol. 2003;  58 311-314
• 7 Wormanns D, Kohl G, Klotz E. et al . Volumetric measurements of pulmonary nodules at multi-row detector CT: in vivo reproducibility.  Eur Radiol. 2004;  14 86-92
• 8 Keil S, Plumhans C, Behrendt F F. et al . Semi-Automated Quantification of Hepatic Lesions in a Phantom.  Invest Radiol. 2008;  44 82-88
• 9 Keil S, Behrendt F F, Stanzel S. et al . Semi-automated measurement of hyperdense, hypodense and heterogeneous hepatic metastasis on standard MDCT slices. Comparison of semi-automated and manual measurement of RECIST and WHO criteria.  Eur Radiol. 2008;  18 2456-2465
• 10 Puesken M, Juergens K U, Edenfeld A. et al . Einfluss des Vaskularisationsgrades auf die automatische Segmentierung und Messung von Lebertumoren nach RECIST in einer biphasischen Multi-Slice-CT (MSCT).  Fortschr Röntgenstr. 2009;  181 67-73
• 11 Bland J M, Altman D G. Statistical methods for assessing agreement between two methods of clinical measurement.  Lancet. 1986;  1 307-310
• 12 Lin X Z, Sun Y N, Liu Y H. et al . Liver volume in patients with or without chronic liver diseases.  Hepatogastroenterology. 1998;  45 1069-1074
• 13 Bland J M, Altman D G. Measurement error and correlation coefficients.  Bmj. 1996;  313 41-42
• 14 Pauls S, Kurschner C, Dharaiya E. et al . Comparison of manual and automated size measurements of lung metastases on MDCT images: potential influence on therapeutic decisions.  Eur J Radiol. 2008;  66 19-26
• 15 Hein P A, Romano V C, Rogalla P. et al . Linear and volume measurements of pulmonary nodules at different CT dose levels – intrascan and interscan analysis.  Fortschr Röntgenstr. 2009;  181 24-31
• 16 Heussel C P, Meier S, Wittelsberger S. et al . Quantitative CT-Verlaufskontrolle von Lebermalignomen nach RECIST und WHO im Vergleich zur Volumetrie.  Fortschr Röntgenstr. 2007;  179 958-964
• 17 Hopper K D, Kasales C J, Van Slyke M A. et al . Analysis of interobserver and intraobserver variability in CT tumor measurements.  Am J Roentgenol. 1996;  167 851-854
• 18 Sohaib S A, Turner B, Hanson J A. et al . CT assessment of tumour response to treatment: comparison of linear, cross-sectional and volumetric measures of tumour size.  Br J Radiol. 2000;  73 1178-1184
• 19 Goei R, Lamers R J, Engelshove H A. et al . Computed tomographic staging of esophageal carcinoma: a study on interobserver variation and correlation with pathological findings.  Eur J Radiol. 1992;  15 40-44
• 20 Padhani A R, Ollivier L. The RECIST (Response Evaluation Criteria in Solid Tumors) criteria: implications for diagnostic radiologists.  Br J Radiol. 2001;  74 983-986
• 21 De Vriendt G, Rigauts H, and M eeus L. A semi-automated program for volume measurement in focal hepatic lesions: a first clinical experience.  J Belge Radiol. 1998;  81 181-183
• 22 Marten K, Auer F, Schmidt S. et al . Automated CT volumetry of pulmonary metastases: the effect of a reduced growth threshold and target lesion number on the reliability of therapy response assessment using RECIST criteria.  Eur Radiol. 2007;  17 2561-2571
• 23 Keil S, Behrendt F F, Stanzel S. et al . RECIST and WHO criteria evaluation of cervical, thoracic and abdominal lymph nodes in patients with malignant lymphoma: manual versus semi-automated measurement on standard MDCT slices.  Fortschr Röntgenstr. 2009;  181 888-895
• 24 Vogel M N, Vonthein R, Schmucker S. et al . Lungenrundherdvolumetrie mit optimiertem Segmentierungsalgorithmus. Genauigkeit bei verschiedenen Schichtdicken verglichen mit ein- und zweidimensionalen Messungen.  Fortschr Röntgenstr. 2008;  180 791-797
• 25 Thiesse P, Ollivier L, Di Stefano-Louineau D. et al . Response rate accuracy in oncology trials: reasons for interobserver variability. Groupe Francais d’Immunotherapie of the Federation Nationale des Centres de Lutte Contre le Cancer.  J Clin Oncol. 1997;  15 3507-3514
• 26 Trillet-Lenoir V, Freyer G, Kaemmerlen P. et al . Assessment of tumour response to chemotherapy for metastatic colorectal cancer: accuracy of the RECIST criteria.  Br J Radiol. 2002;  75 903-908
• 27 Watanabe H, Yamamoto S, Kunitoh H. et al . Tumor response to chemotherapy: the validity and reproducibility of RECIST guidelines in NSCLC patients.  Cancer Sci. 2003;  94 1015-1020
• 28 Park J O, Lee S I, Song S Y. et al . Measuring response in solid tumors: comparison of RECIST and WHO response criteria.  Jpn J Clin Oncol. 2003;  33 533-537

1 In a new software version (release 5.5) the options for a segmentation presetting were reduced and the seed-point method will now be started by a rough line marking the lesion. Both modifications are supposed to reduce the interobserver and intraobserver variability of semi-automated lesion measurements. This version will be released in late 2009.