We would like to thank ShahidSales and colleagues for their thoughtful letter about our article (1). They raised three important methodologic issues: (i) the isolation of a relatively pure population of tumor cells; (ii) the choice of the housekeeping gene (HKG) for accurate normalization; and (iii) the reevaluation and validation of the IHC score. We appreciate their interest in our study about the role of SFRP1 in nasopharyngeal carcinoma (NPC) metastasis and agree that these important considerations must be addressed.
NPC tissue is characterized by cellular heterogeneity. Therefore, the isolation of a relatively pure population of tumor cells may allow the identification of tumor-specific molecular alterations. Laser-capture microdissection is a useful technique that allows efficient and precise isolation of pure cell populations or even of single cells from complex heterogeneous tissue structures (2). However, tissues from NPC biopsies are usually low in volume and comprise numerous cancer nests separated by stroma. These features greatly hinder laser-capture microdissection. Recent studies, based upon hematoxylin and eosin staining, have reported an alternative method to decrease the bias caused by inclusion of normal tissue—samples are selected in which tumor tissue comprises the major area for analysis (3). In our study, we performed IHC and determined the SFRP1 protein expression by evaluating the staining intensity and the proportion of positive cells in tumor tissue alone.
Accurate normalization is a crucial step for gene expression studies. To this end, the publication in 2009 of the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) established guidelines, which recommend that at least two validated reference genes be used for accurate normalization. Subsequently, it became increasingly apparent that a prior test of reference gene stability is also necessary before the genes are chosen. In the past few years, there appears to be a trend toward testing against more reference genes. However, GAPDH is still commonly chosen as a reference gene and has been since it became a standard in the 1980s and early 1990s during the era of Northern blotting. Chapman and colleagues reported that over 72% of studies still use GAPDH, ACTB, or 18S rRNA as single normalizing genes (4). GAPDH, which is a housekeeping gene, has also been widely used to normalize the expression of genes in studies of NPC. With developments in methodology, the geometric means of multiple internal-control genes and the determination of gene stability appear to be setting trends in gene expression studies.
Today, distant metastasis remains the main reason for treatment failure in NPC. Molecular prognostic markers are urgently needed that can predict distant metastasis and guide individualized treatment for patients with NPC. Recently, we reported that a 6-hypermethylated gene panel including SFRP1 was associated with poor survival in patients with NPC (5). It has been reported that SFRP1, which is hypermethylated, can help predict prognosis in various cancers. These findings indicate that SFRP1 should probably be a therapeutic target for cancer therapy. However, whether SFRP1 plays a direct role in NPC remains unclear. IHC is an imprecise and empirical method with which to examine the expression of SFRP1 in NPC patients. This is because the IHC result depends on the antibody used and pathologists' expertise. We chose an antibody that had been widely used in other immunohistochemical staining studies, and all tumor samples were evaluated and reevaluated by at least two independent pathologists.
Although there are some imperfections, the experimental techniques and the methods we used to analyze our data have been widely used and verified by our peers. Nevertheless, with the development of new methodologies, the design of experiments will require modification in the future.
See the original Letter to the Editor, p. 415
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.