Plasma DNA Microsatellites as Tumor-specific Markers and Indicators of Tumor Progression in Melanoma Patients

INTRODUCTION

Recent advances in tumor genetics have revealed that genesis and progression of tumors follow an accumulation of multiple genetic alterations, including inactivation of tumor suppressor genes and/or activation of proto-oncogenes (1) . Frequent LOH 3 of DNA microsatellites on specific chromosomal regions have been reported in various types of malignancies. The frequency of LOH in tumors, along with homozygous deletions at specific chromosome sites in tumor cells, suggests the involvement of putative tumor suppressor genes or oncoproteins related to tumorigenesis and/or tumor progression. LOH analysis, combined with genetic linkage analysis on pedigrees of familial cancer (2, 3, 4, 5, 6) or homozygous deletion analysis (7, 8, 9) , has identified candidate tumor suppressor genes. Allelic losses or microsatellite alterations on specific chromosomes are the most common genetic alterations observed in a wide variety of malignancies (10, 11, 12, 13, 14, 15) . Recently, LOH analysis of tumor tissues has been shown to be of prognostic value (10 , 14 , 16) . To date, however, microsatellite analysis for LOH has been primarily carried out using labor intensive assessment of archival paraffin tumor sections.

There is evidence that naked DNA is released, enriched, and remains stable in the blood of cancer patients (17, 18) . Recently, tumor-specific DNA has been detected in the plasma and serum of lung, head and neck, and colon cancer patients (19, 20, 21, 22, 23) . This suggests that cell-free plasma/serum is a source for detecting cancer-specific DNA markers. In the past, tumor-associated markers such as proteins/glycoproteins have been used for diagnosis or prognosis of progression in patients. However, the specificity of these assays is limited because the majority of these markers are not tumor-specific and are found in normal cells. To date, tumor-specific genetic markers have been assessed primarily in tumor biopsies. However, in advanced-staged patients, surgery is not always performed, which limits the availability of tumor tissue for genetic assessment. The detection of tumor-specific genetic markers in cancer patients at distant sites from the tumor, such as in the blood, provides a unique and valuable tumor genetic marker assay for diagnosis and prognosis.

Melanoma can be a highly aggressive cancer that becomes very difficult to manage clinically when disease progression occurs. Patients diagnosed with early-stage primary melanoma (AJCC stages I and II) who undergo surgical treatment have a low incidence of disease recurrence and usually a positive prognosis (24, 25, 26) . However, when recurrence of disease occurs, it is often difficult to manage. There is considerable variability in the extent and type of disease progression in AJCC stage III and stage IV patients (25 , 26) . Therefore, the genetic analysis of recurrence and stages of disease spread may aid in improving prediction of disease progression and in determining the most effective strategy for treatment. The stepwise biological progression of melanomas from benign nevi to malignancy has been documented (27) . However, the corresponding genetic analysis of individual stages of tumor progression is limited. The detection and understanding of genetic changes relating to melanoma progression is not well understood, particularly during disease progression from AJCC stage II to stage IV.

In melanoma, deletions and mutations of several known tumor suppressor genes, such as TP53 and CDKN2, have been reported; however, they occur at a low frequency (28, 29, 30) . The tumor suppressor gene CDKN2 at 9p21 encoding p16^ink4 protein is associated with sporadic and familial melanoma (29, 30, 31) . There are frequent homozygous deletions and LOH at the microsatellite loci 9p21 region in melanoma. However, the absence or mutation of CDKN2 has not been found frequently or well correlated with tumor progression. Additionally, there are other potential promising DNA markers (such as microsatellites) with frequent LOH on chromosome loci 1p36, 3p25, 6q22-q26, 10q24-q26, and 11q23 that have been reported in melanomas (13 , 16 , 32, 33, 34, 35) .

In this study, we examined the plasma of 57 advanced and 19 early clinically-staged melanoma patients using a panel of 10 microsatellite markers representing six chromosomal regions. Of these patients, 40 with matched tumor lesions available were assessed. The study demonstrated that multiple LOH markers can be detected in the plasma of melanoma patients and not in healthy donors. The study also demonstrated that melanomas release tumor-specific genetic markers in blood that highly correlate to the patients’ respective melanoma lesion. There was a significant correlation between frequency and combinations of specific microsatellite markers with LOH to clinical disease progression.

MATERIALS AND METHODS

Specimens.

Blood (5 ml) was collected in sodium citrate-containing tubes (Becton Dickinson, Franklin Lakes, NJ) from a total of 76 patients diagnosed with melanoma at the John Wayne Cancer Clinic. Similarly, blood was drawn from 20 healthy donor volunteers. The plasma was immediately separated from blood cells by differential centrifugation at 1000 × g for 15 min, filtered through a 13-mm serum filter (Fisher Scientific, Pittsburgh, PA) to remove any potential cells, and then cryopreserved at −30°C until DNA extraction. Blood was spotted on FTA blood stain cards (Fitzco, Minneapolis, MN) for normal genomic DNA extraction, as well as long-term storage. Respective WBCs from individual patients were used as normal DNA controls. Cells were separated by differential centrifugation from RBCs using Puregene RBC lysis solution (Gentra Systems, Minneapolis, MN). The cell pellet was then washed with PBS. Corresponding tumor tissues were microdissected from two to three 10-μm serial sections of formalin-fixed paraffin-embedded blocks, as previously described (36) . All microdissected tissue sections were identified histopathology positive for malignant melanoma cells (28) .

DNA Isolation.

Control lymphocyte DNA was isolated using DNAzol (Molecular Research Center Inc., Cincinnati, OH). In brief, cell pellets were homogenized with 1 ml of DNAzol and precipitated by the addition of 0.5 ml of 100% ethanol. After centrifugation, precipitated DNA was then washed twice with 95% ethanol and resuspended in 10 mm Tris (pH 8)-1 mm EDTA buffer. Plasma (1 ml) was diluted at 60% with a solution of 0.9 M NaCl, 1% SDS, and proteinase K. The diluted plasma was shaken and incubated at 37°C overnight in the presence of an equal volume of phenol-chloroform isoamyl alcohol (25:24:1). After centrifugation for 15 min at 1000 × g, the aqueous phase was collected, extracted with an equal volume of phenol-chloroform isoamyl alcohol, and precipitated by isopropanol. Tumor lesions microdissected from 40 paraffin-embedded tissue blocks were incubated with xylenes at 37°C overnight (12) . The pellet was recovered after centrifugation and washed twice with 1 ml of 100% ethanol. The remaining material was dried by vacuum centrifugation, incubated with proteinase K in lysis buffer (50 mm Tris-HCl, 1 mm EDTA, and 0.5% Tween 20) at 37°C overnight and then boiled at 95°C for 10 min in a heat block.

Microsatellite Markers and PCR.

Ten primer sets for PCR amplification of microsatellite markers were chosen on six chromosome arms. The following CA_(n)-repeat microsatellite markers were used: D1S214 at 1p36.3; D1S228 at 1p36; D3S1293 at 3p-3p24.2; D6S264 at 6q25.2–6q27; IGFIIR at 6q25-q27; D9S157 at 9p23-p22; D9S161 at 9p21; D10S212 at 10q26.12–10q26.13; D10S216 at 10q24-q26; and D11S925 at 11q23.3–11q24. Primer sets for PCR were obtained from Research Genetics, Inc. (Huntsville, AL), and sense primers were labeled with a fluorescent FAM or Cy5 dye. Genomic DNA (∼50 ng) was amplified by PCR in 25-μl reactions containing 1 × PCR buffer (6.7 mm Tris, 16.6 mm ammonium sulfate, 6.7 mm EDTA, and 10 mm β-mercaptoethanol), 6 pmol of each primer, 1 unit of Taq DNA polymerase, 0.8 mm of each dNTP, and 1.5 mm MgCl₂. Forty PCR cycles were performed, with each cycle consisting of 30 s at 94°C, 30 s at 50–56°C, and 90 s at 72°C, followed by a final extension step of 72°C for 5 min.

LOH Analysis.

PCR product (5 μl) mixed with 2 μl of loading dye (100% formamide, 2 mm EDTA, and 2% dextrane blue) was incubated at 95°C for 15 min. The concentrated samples were electrophoresed on a 6% denaturing PAGE containing 7.7 M urea at 1600V for 2 h. The fluorescent-labeled PCR product images were scanned by a fluorescent/optical Genomyx SC scanner (Genomyx Corp., Foster City, CA). After the image acquisition was completed, image files were analyzed by Adobe Photoshop software (Adobe, San Jose, CA). Densitometry analysis was performed with the imaging software ClaritySC (Media Cybernetics, Silver Spring, MD). Intensity calculations and intensity comparison of the specific alleles in lymphocytes, plasma, and tumor DNA were performed to evaluate for LOH. Tumor and plasma were scored as exhibiting LOH if there was an absence or more than a 50% reduction in the intensity of one allele compared with the respective allele in the normal matched lymphocytes.

Statistics.

Correlation between patients’ matched tumor and plasma samples for individual microsatellite markers were assessed using the Kappa agreement test. Logistic regression was used to test the association between the number of positive markers and AJCC Stage (37) . Spearman correlation was estimated to evaluate the association between the number of positive markers, Breslow’s thickness, and Clark’s level. The χ² test was also used when data were cross-classified into a contingency table.

RESULTS

LOH Analysis in Plasma and Melanoma Biopsies.

LOH was initially assessed at 10 different loci on six different chromosomes in paired tumor biopsies and plasma from 40 melanoma patients. DNA was extracted and detected from the plasma, tumor biopsies, and lymphocytes of all patients. The frequency of LOH varied between 23% and 53% in tumor biopsies and 4% and 33% in plasma for individual microsatellite markers in informative cases. For informative melanoma tumors (Table 1 ⇓ ; Fig. 1 ⇓ ), the most frequent microsatellite markers with LOH were detected at loci D6S264, D9S161, D10S216, and D11S925, respectively. Additionally, in informative plasmas from melanoma patients (Table 1 ⇓ ; Fig. 1 ⇓ ), the most frequent microsatellite markers with LOH were detected at loci D10S216, D3S1293, D6S264, and D1S214, respectively. Microsatellite markers D6S264 and D10S216 were among those loci that frequently had LOH for both plasma and tumor. The least frequent microsatellite marker for LOH detection was IGFIIR, in both plasma and tumor. Representative LOH in tumor and plasma as compared with lymphocyte DNA is shown in Fig. 2 ⇓ . LOH was not observed for any of the microsatellite markers tested on the lymphocytes and plasma DNA from 20 healthy donors.

A summary of LOH in tumor and plasma from 40 melanoma patients using a panel of 10 microsatellite markers is shown in Table 2 ⇓ . LOH for at least one microsatellite marker was shown in tumor samples of 34 of 40 (85%) patients and in plasma samples of 23 of 40 (58%) patients. Overall, a significant correlation (P < 0.0001) between LOH matched-paired plasma and tumor specimens of each individual microsatellite marker was shown. Twenty-one of 23 (91%) patients with LOH markers in plasma showed LOH in their respective tumors, when assessing all microsatellite markers. Two of these patients had LOH markers in plasma and not tumor. In 13 patients, LOH was observed in tumor and not plasma. In one interesting case, the plasma DNA showed LOH at D3S1293 but not at D9S157, whereas the tumor sample of this patient showed LOH at both loci. These differences may be due to different metastases at various sites or clonal heterogeneity within the tumor, in which there are tumor cell colonies with or without specific microsatellite marker(s) LOH.

Plasma LOH Correlation with Different Clinical Stages of Melanoma.

Plasma was assessed from 76 melanoma patients with different clinical stages (AJCC) of disease: 7 stage I; 12 stage II; 30 stage III; and 27 stage IV. The mean age of the patients was 51.6 ± 17.0 SD, consisting of 28 females and 48 males. The mean Breslow’s thickness was 2.07 mm. The majority of the patients had a Clark’s level of III or IV. In informative patient cases, a correlation between the frequency of plasma LOH and AJCC stage was performed (Table 3) ⇓ . The frequency of LOH was higher in more advanced stages of disease: 5 of 19 (26%) patients with stages I and II had LOH, whereas 33 of 57 (58%) patients with stages III and IV had LOH in at least one locus. One interesting finding was that in a stage I patient’s plasma and primary tumor, LOH detected at loci D3S1293 and D10S212 showed a different pattern of allele loss in the tumor and plasma (Fig. 3) ⇓ . The longer allele was deleted in both plasma and tumor for D3S1293, whereas for D10S212, LOH was at the longer allele in the plasma and at the shorter allele in the tumor. The discrepancy may be due to the heterogeneity of the primary tumor or the presence of subclinical metastasis. Clinical follow-up for disease recurrence may help in determining the cause of these events.

Statistical analyses were performed to determine whether the frequency of LOH in plasma correlated to clinical stage and known melanoma prognostic factors. Assessment of clinical correlation between stage and individual microsatellite markers showing LOH in the patients’ plasma was performed. Overall, there was a significant correlation between the number of LOH microsatellite markers detected within a patient’s plasma and AJCC stage in the 76 patients (P = 0.02). Only at loci D3S1293 was there a significant correlation (P = 0.02) between LOH detection and clinical progression of disease. The next closest correlation with an individual marker was at D1S228 (P = 0.065). In the correlation between the progression of different clinical stages of disease and microsatellite marker combinations showing LOH, the combinations of D9S157 and D3S1293 (P = 0.01), D9S157 and D1S228 (P = 0.05), and D11S925 and D3S1293 (P = 0.01) were the most significant. These correlations were independent of known prognostic factors for melanoma. The microsatellite marker with LOH combination of D1S228 and D3S1293 showed a trend toward significance (P = 0.07).

We also compared LOH in plasma DNA with other clinicopathological parameters and clinical status of disease (no evidence of disease or alive with disease) at the time of blood collection. There was no significant correlation between the frequency of LOH in plasma or tumor, and standard prognostic factors such as Breslow’s thickness or Clark’s level. Patients were also assessed for correlation of LOH presence with recurrence or progression of disease (Table 2) ⇓ , and because the mean follow-up of patients is only about 1 year, the results cannot be appropriately evaluated.

DISCUSSION

The development of molecular markers is needed to improve diagnosis and prognosis of disease and to assess tumor progression in melanoma patients. Better profiles of tumor genetic changes are needed to determine mechanics associated with disease progression. We and others have examined molecular markers such as melanoma-associated antigen mRNA markers in RT-PCR assays to detect melanoma cells in blood, lymph nodes, and various other organs (38 , 39) . These molecular markers have been shown to be very useful in detecting metastatic melanoma cells and disease progression. Although these RT-PCR-based markers have a high level of sensitivity, there are still some limitations, such as potential false positives and the expression of mRNA markers by normal cells. Other limitations in conducting RT-PCR analysis for large clinical studies are the high potential of RNA contamination and integrity of mRNA available. The mRNA expression level of a particular tumor can vary among tumor cells, which also reduces the sensitivity of the assay. However, from these studies, we have determined that multiple molecular markers are needed in assessment of melanoma progression. This approach takes into consideration the heterogeneity of tumors and the dynamics of tumor genetic changes occurring during tumor progression.

Analysis of DNA markers for somatic mutations of tumor suppressor genes and oncogenes and chromosome alterations are more specific to tumors and require less logistical stringency than mRNA analysis. The frequency of known tumor suppressor genes or oncogenes with mutations in melanoma is <25% individually, which makes them unsuitable as molecular markers for detection overall. DNA mutations of the K-Ras gene, previously reported (Y. F.) to be present in the plasma of patients with colorectal and pancreas cancer, has been correlated to clinical status (22) . However, this molecular marker occurs quite frequently in these cancers, but not in melanoma. DNA cancer-cell specific markers do not have many of the problems associated with mRNA markers. In melanoma, loss of loci on specific chromosome arms is one of the most common genetic events occurring in these tumors. The correlation of LOH markers to clinical progression or prognosis has been shown in other cancers, but has not been well described for melanoma. The assessment of a combination of microsatellite markers with high frequency of LOH as genetic markers in blood may be very useful for monitoring melanoma progression at different stages of disease. Blood is logistically practical for the monitoring of multiple patients sampled at different time points during tumor progression. Traditional approaches using embedded tissue for analysis of LOH are not always practical and, as a result, there are major gaps of information missing on events during tumor progression.

The panel of 10 microsatellite markers used in the study showed LOH with at least one marker in 85% of the melanoma tumors assessed. Twenty-three of 34 (68%) patients who showed LOH for at least one marker in tumor also showed LOH in plasma. These analyses of matched patients’ plasma and tumor were highly statistically concordant for all microsatellite markers studied. The assay is very specific in that none of the normal samples tested showed LOH at any loci. The frequency and combinations of LOH at specific microsatellite loci can vary considerably among tumors with different histological origin. These results demonstrate that tumor-specific DNA is released frequently into the blood circulation and remains stable in melanoma patients. In contrast to mRNA, specific tumor-derived DNA is more stable in blood and is more resistant to rapid degradation. Previous studies revealed that the serum of cancer patients contains approximately two to four times the amount of free DNA as that of normal donors (23 , 40) . Future studies will help determine the half-life of these microsatellite markers and whether specific markers stay longer in blood circulation (i.e., resistant to degradation). The remarkable finding is that the LOH of the individual microsatellite markers can be detected easily in the plasma. The mechanism of how DNA (LOH marker) is released into the blood circulation is unknown. This can be related to cell death and necrosis in tumors and/or destruction of tumor cells circulating in blood. A significant correlation was observed within individual patients who showed LOH in plasma and paired tumor microsatellite markers. This demonstrated the relative detection sensitivity from a single bleed. Studies on several bleed samples over several months from the same patient with LOH in plasma have shown consistency in presence of the marker.

The frequency and the number of microsatellite markers with LOH in plasma significantly increased in more advanced clinical stages of disease. These findings support previous work on melanoma and other cancers that show that as tumors progress, the number of genetic changes accumulates. Only at loci D3S1293 was LOH significantly correlated to disease progression. Located at this site are the tumor suppressor gene VHL (3p25-p26) and other genetic abnormalities that are frequently found in renal cell carcinomas (41) . Interestingly, two of the most frequently and well studied microsatellite loci in melanoma, 9p21 and 6q, did not correlate with disease progression individually. This may be due to the limited number of patients analyzed in the study. LOH of these two markers has been detected at early stages of melanoma development therefore, the loci alone may not be significantly correlated with later stages of tumor progression. Combinations of specific LOH markers may be necessary for tumor progression to be successful. One of the major problems in interpretations of these analyses is that tumor progression clinicopathology at later stages is highly variable. A recent study on LOH of microsatellite marker at 6q of melanomas showed a correlation with poorer clinical outcome (16) . The combinations of genetic changes are more likely to be correlative to disease progression. In our study, LOH at 9p combined with LOH at 3p or 1p showed significant correlation. Assessment of LOH at 3p alone or in combination showed better correlation with disease progression than any other marker.

This study illustrates the clinical se of microsatellite analysis in detecting tumor DNA in plasma of melanoma patients. The analysis of LOH in plasma provides a logistically practical assay to monitor genetic changes during melanoma progression. The study demonstrates that at early clinical stages, release of DNA (LOH marker) is limited. Plasma LOH analysis may be more suitable to monitor stage II to stage IV progression before and during therapy as well as during posttreatment follow-up. The markers may be also useful to detect subclinical disease recurrence in disease-free patients. Tumor progression is dynamic, and the genetic changes that concurrently occur are also ongoing. The most significant advantage of this approach in assessing plasma compared with direct analysis of tumor biopsies is the ability to monitor disease progression and genetic changes without assessing the tumor. This is particularly important during early phases of distant disease spread in which subclinical disease is undetectable by conventional imaging techniques. Additional retrospective and prospective analyses are being carried out to determine the prognostic significance of the DNA microsatellite markers during treatment and as overall predictors of disease outcome.