This study presents the results obtained from analyzing protein extracts obtained by a novel method by SELDI-TOF MS. We have shown that water-soluble proteins sampled from punch biopsies are promising to assist the diagnosis of normal and CIN2-3.
We found that 24 h immersion of cervical punch biopsies in RPMI at 4°C can be used to harvest 1 mg/ml water soluble proteins without compromising the routine immunohistochemical analysis of the tissue. Moreover, the water soluble proteins from the biopsies give good discrimination between samples with normal and CIN2-3 epithelia.
SELDI-TOF MS technology has been applied for several years to address the need for new markers for early cancer detection [19
] or for different lymphomas [25
]. Most studies have compared diseased and healthy persons and have been useful for a better understanding of the development of several types of cancer. However, it is more difficult to develop biomarkers which can discriminate between the different development stages of a neoplasia as the differences are much smaller than between healthy and diseased persons. Yet, the model presented here shows that this still is possible with the technology used. CIN represents the first step in the development of cervical cancer through the three different CIN phases. In a future study, it is important to analyze whether SELDI-TOF MS or other mass spectrometric techniques can discriminate between CINs that will regress spontaneously from CINs that will persist or progress to invasive cancer. Furthermore, the protein collection method that was developed is of great interest for studying not only cervical neoplasia but it could also be applied to other organ tracts. The method of placing a biopsy in a cell culture medium for 24 h at 4°C is considerably longer than 1 h in PBS at 37°C used by Celis et al. [21
], and enables the collection of proteins from a biopsy without interfering with the essential diagnostic information. The three peaks found to separate normal tissue and CIN2-3 lesions did not have the same m/z-values as those reported found from SELDI-TOF MS analysis of cell lysates of normal and invasive cervical cancer [26
] or from plasma samples of CIN patients [27
]. The protein concentrations in the samples were comparable to those found in samples from mucosal collection using sponges [28
]. In addition, the RPMI immersion method has additional value to conventional FFPE analysis as it enables the collection and analysis of water soluble proteins which are not available from formalin-fixed paraffin-embedded tissue protein extracts. Proteins like cytokines and chemokines are collected in a volume enabling the analysis by different techniques like SELDI-TOF MS, LC-MS/MS or by ultrasensitive Elisa techniques like electrochemiluminescence.
The reviews published by Dijkstra et al. [29
] and DeBock et al. [30
] discuss several sources of variation that needs to be addressed. An essential aspect of a new laboratory test with possible diagnostic and therapeutic implications is the variation sources in all steps of the analysis [31
]. These include: 1. Pre-analytical (a. each time getting the same samples, by using the same biopsy device; b. taking two samples from the same patient and put them in two different RPMI tubes; c. standardize the incubation conditions and time)
; 2. Analytical part (a. binding and cleanup on the SELDI-chip; b. reproducibility of the SELDI-TOF measurement of the same sample; c. take two samples from the same RPMI sample and test the reproducibility); 3. Post-analytical (a. peak processing and clustering; b. data analysis).
As to sampling device, we always used the same biopsy forceps. The samples were all taken by the same experienced gynaecologist. We also analyzed two nearby samples from 5 consecutive patients and compared the results. Two neighbouring biopsy samples in the cervix from the same patient also gave comparable results. We standardized the incubation conditions (RPMI, 4oC) and time (24 h).
As to the analytical part, all the technical aspects of the Seldi chip were standardized as described. We also tested the reproducibility of the same samples (see Fig. ). Moreover, two different technicians analyzed the same samples, with the same results.
As to peak processing, clustering and data analysis, we used strictly protocolized standard operation procedures.
The results are promising but should be confirmed in a larger set of independent samples. In an ongoing study not published yet, a subset of supernatants has been characterized using LC-MS/MS. Further work to identify the candidate biomarkers found using this model will be done using LC-MS/MS (one- or multidimensional [32
]) of tryptic digests after fractionation by chromatography or two-dimensional gel electrophoresis [34
]. An identification of the possible biomarker candidates found using this model will possibly give new insight to the mechanisms related to the Human Papilloma Virus infection.