Recently, in vitro anticancer drug sensitivity tests using clinical specimens have been used to provide data for designing individualized chemotherapies. Several in vitro anticancer drug sensitivity tests have been developed for various types of malignant tumors, and these tests have been applied experimentally as well as clinically (11-14). The collagen gel droplet embedded culture drug test (CD-DST) is an in vitro anticancer drug sensitivity test (15-17) that has been used at our institute in chemotherapy for patients with non-small cell lung cancer (NSCLC) (18-20) as well as those with other thoracic tumors (21,22). So far, this test has been used to assess surgically resected specimens from NSCLC primary lesions and to provide data regarding their sensitivity to anticancer drugs and has also been clinically applied to aid the development of individualized chemotherapies for NSCLC patients who have suffered postoperative recurrence (18,20). In fact, good predictability was obtained when the test was used to aid the treatment of recurrent disease, and the accuracy of treatment response predictions based on the CD-DST data was as high as 70%, but this was still not satisfactory because these chemosensitivity data were obtained from primary NSCLC tissues, not systemic metastatic tissues (20).

The clinicopathological data of the patients enrolled in the present study are summarized in Table 1. The pathological stage (p-stage) of the disease was based on the general guidelines of the Japan Lung Cancer Society (23). As described above, the patients were divided into two groups, those (n=26) with CDDST data for primary and nodal metastatic lesions and those (n=6) with CD-DST data for primary and distant metastatic lesions. The former group consisted of 8 squamous cell carcinomas, 14 adenocarcinomas, 3 large cell carcinomas, and one adenosquamous cell carcinoma. The nodal specimens used for the test were obtained from dissected mediastinal or hilar lymph nodes while the metastases were histologically confirmed during surgery. The latter group consisted of 1 squamous cell carcinoma, 4 adenocarcinomas, and 1 large cell carcinoma. The metastatic lesions were diagnosed using intraoperative and postoperative histological examinations.

CD-DST was performed as described previously by Kobayashi et al. (15-17). In brief, each surgically obtained specimen was finely minced using a scalpel and digested in cell dispersion enzyme solution (EZ, Nitta Gelatin Inc., Osaka, Japan) for 2 hr. The dispersed cancer cells were then washed twice, collected by centrifugation at 250g for 3 min, filtered through an 80 um nylon mesh, and then incubated in a collagen gel coated flask (CGflask, Nitta Gelatin Inc.,) in a CO₂ incubator at 37 ℃ for 24 hr. Only the viable cells adhering to the collagen gel were collected and suspended in the reconstructed type I collagen solution (Cellmatrix Type CD, Nitta Gelatin Inc.) at a final density of 1x10⁵ cells/ml. Three drops of the collagen-cell mixture (30ul/drop) were placed in each well of a 6-well multiplate and a 60 mm dish and allowed to gel at 37℃ in a CO₂ incubator for 1 hour. The final concentration was about 3x10³ cells/collagen gel droplet. The culture medium was overlaid on each well, and the plate was incubated in a CO₂ incubator at 37℃ overnight.

In addition to our series (18,20), there have been several reports regarding the clinical application of in vitro sensitivity tests for the treatment of lung cancer patients. Kawamura et al. (19) described the survival benefit of CD-DST-based chemotherapy for patients with stage IV lung cancer. Yoshimasu et al. (24) also reported the usefulness of another in vitro chemosensitivity test, the histoculture drug response assay (HDRA), for treating postoperative recurrence in lung cancer patients. Recently, Tanahashi et al. (25) reported the clinical application of the HDRA for postoperative adjuvant chemotherapy in lung cancer patients and demonstrated that overall survival was prolonged by treatment using an HDRA-sensitive regimen. In addition, there have also been some promising reports regarding other novel chemosensitivity tests for the treatment of patients with NSCLC (10,26). In particular, such an in vivo test system as patient-derived xenograft model described by Dong et al. (10) was newly promising for predicting drug sensitivities. Thus, it is considered that these chemosensitivity tests may be clinically applicable for sensitivity test-guided, individualized treatment of cancer patients. However, it has been well recognized that there are some limitations to apply these in vitro tests in clinical practice enough. In fact, there are still some technical problems of primary culture failure, anticancer drug level, bacterial contamination, measurement only for cancer cells, and so on. Anyway, these tests including CD-DST (15-17) have been developed while overcoming such technical problems step by step.

Interestingly, we must also pay particular attention to the fact that most of these analyses are based on sensitivity data obtained from primary, not metastatic, lesions. In other words, it is possible that these data do not reflect the characteristics of all tumor tissues in a particular patient. Since chemosensitivity data could not be obtained for all sites, chemotherapy was performed based on the data of the most representative primary site in patients with NSCLC (18,20,24). However, the prediction of chemotherapeutic effects using sensitivity tests was not always satisfactory in our series (18,20) or those of others (19,24). Unfortunately, the reason for these problems is unclear. From this standpoint, this study is extremely important for elucidating the cause of predictive failure.

According to the present study, although a minority of patients had metastatic lesions that were more in vitro-sensitive than the primary lesion, some anticancer drugs, for example, TXT, VNR, and GEM, showed significantly less in vitro-sensitivity in metastatic lesions than in primary lesions, and then, these differences resulted in a lower incidence of in vitro-sensitive cases for TXT and VNR. In contrast, only small differences in sensitivity were detected between the primary and metastatic lesions for CDDP and 5-Fu. Furukawa et al. (27) reported similar results for various anticancer drugs using specimens from primary breast cancer lesions and their paired nodal metastatic tumors by the HDRA. Interestingly, they also showed that the sensitivity of the metastatic nodal lesions to CDDP was not different from that of the primary lesions. Therefore, when performing individualized chemotherapy based on CD-DST data using primary tumor specimens, false-sensitive regimens, including some anticancer drugs that display in vitro-sensitivity, e.g., TXT and VNR, might be selected.

Surprisingly, the effectiveness of some anticancer drugs depended on the metastatic route or site. GEM and CDDP showed a trend towards reduced sensitivity in non-lymphatic metastatic lesions compared with lymph node metastases, although the observation was not subjected to statistical analysis because of the small number of samples. This trend might be closely associated with our previously reported result (20); i.e., in CD-DST-based chemotherapy for recurrent disease, the predictive accuracy of CD-DST data was highest for lymph node recurrence. In contrast, among patients with pleural or bone metastasis, few such associations between CD-DST data and response were observed, despite the fact that we performed chemotherapy with an in vitro- sensitive regimen (20). In addition, a similar tendency was also found in a recent report: Tanahashi et al. (25) described that the incidence of postoperative lymph node recurrence was low in lung cancer patients undergoing adjuvant chemotherapy involving an in vitro-sensitive regimen. Thus, for some anticancer drugs, the metastatic route or site may also influence the predictive performance of CD-DST.