We read with great interest the letter by Kuo et al regarding our use of a cocktail of cytokeratin (CK) antibodies in our Cancer article.1 We respectfully disagree with the authors' assertion that we overlooked the manufacturer's information and used redundant anti-CAM5.2 and anti-CK8 antibodies and that we may be misleading readers who may be designing a cost-effective CK antibody cocktail.
We used a cocktail of CK antibodies including AE1/AE3 (Dako, Carpinteria, Calif), CAM5.2 (Becton Dickinson, San Jose, Calif), MNF116 (Dako), and CK8/CK18 (Zymed Laboratories, San Francisco, Calif). We agree with the authors that MNF116 and CK8/CK18 can identify CKs that are not recognized by AE1/AE3, including CKs such as CK17 and CK18. We did not use individual CK8 (Invitrogen, Carlsbad, Calif) or CK18 (Invitrogen) antibodies as referred to by the authors but used the CK8/CK18 antibody (Zymed Laboratories). We agree with the authors that the antigenic epitopes recognized by the CAM5.2 antibody, which mainly recognizes CK8 with weaker reactivity against CK7, can also be recognized by the CK8/CK18 antibody and the AE3 component of the AE1/AE3 antibody. However, it is to be noted that the affinity of the antibodies, which were produced differently, against a common epitope can vary, which can result in mild if not marked variability in their sensitivities with respect to epitope recognition. The variability in the sensitivity of these 2 antibodies has been reported in earlier studies that compared these 2 antibodies directly.2 We have provided some recent reports in the pathology literature in which AE1/AE3 and CAM5.2 were used for the investigation of undifferentiated and metaplastic carcinomas.3, 4 In essence, we used a very robust cocktail of CK antibodies with a wide spectrum of specificity to increase the sensitivity of detection of disseminated tumor cells in the bone marrow of patients with breast cancer.