Problems in radiographic estimation of response to chemotherapy and radiotherapy in small cell lung cancer.

Assessment of the response to therapy in small cell lung cancer requires serial tumor measurements with chest radiographs before and after treatment. Those lesions that are not measurable may be evaluated for response using subjective criteria. We studied interobserver variability in tumor measurement in 21 patients with small cell lung cancer. In addition, we analyzed the effect that the variability in measurement had on the estimation of response to combination chemotherapy with and without chest radiotherapy. Half the readers agreed that pretreatment radiographs were measurable. Posttreatment, they were more often unmeasurable. Starting from a base of 100%, representing all measurable pretreatment films, posttreatment measurability rates fell to 78% and 53% on the 2-month and 4-month films. After radiotherapy, changes particularly reduced the readers' ability to measure the tumors. There was also less interobserver agreement on response after radiotherapy. The intraclass correlation coefficient for partial or complete response versus no response was 0.42 where no chest radiotherapy had been administered and 0.17 after radiotherapy. A measurable lesion on the initial radiograph was important in improving the consistency of evaluation of complete versus partial response particularly in films taken after chemotherapy alone. These data indicate that there is only fair agreement as to whether tumors were measurable and whether a response had occurred. More consistent results were seen in films taken before administration of radiotherapy. A measurable pretreatment film was important in improving interobserver consistency in differentiating a complete from a partial response.