K-ras gene mutations in the diagnosis of fine-needle aspirates of pancreatic masses: prospective study using two techniques with different detection limits

Abstract

Detection of K-ras mutations may be useful in the evaluation of pancreatic cancer. The aim of this study was to assess, in a prospective design, the diagnostic utility of K-ras mutation analysis in 62 consecutive fine-needle aspirates of pancreatic masses, using two PCR-based techniques—standard and enriched—with detection limits of a mutant allele in the presence of 10² or 10³ wild-type alleles, respectively. Cytology alone offered a diagnostic sensitivity of 75%. The enriched higher sensitivity detection technique, in combination with cytology, offered a diagnostic sensitivity of 91% without false positives. The molecular analysis would have contributed to diagnosis in an additional 14 cases of pancreatic cancer. The standard technique contributed to diagnosis in an additional 9 cases. These results strongly support the use of the enriched method of detecting K-ras mutations as a complement to cytology in the evaluation of pancreatic masses.

Adenocarcinoma of the exocrine pancreas is the fourth leading cause of cancer death in Western countries (1). In Catalunya, Spain, its incidence is 8 cases per 100 000 inhabitants per year, and it has increased progressively over the last decades (2). Despite the poor prognosis, patients with localized disease may be cured with surgery (3); however, it is still difficult to diagnose pancreatic cancer in the earlier stages. When pancreatic cancer is clinically suspected and a pancreatic mass identified by ultrasonography or computed tomography scan, a guided percutaneous fine-needle aspirate (FNA)1 can be obtained; this may be the only sample available for diagnosis in most patients (4)(5).

The high incidence of mutations at codon 12 of the K-ras gene (65–100%) (6) leads one to consider them as a potential tumor marker at the tissue level. The development of PCR-based techniques for detection of K-ras mutations has allowed its use in the clinical setting. Data suggest that a combination of cytological examination and K-ras mutation detection in cellular material may improve diagnostic accuracy (7)(8)(9)(10). However, K-ras mutations have been detected not only in intraductal carcinomas but also in pancreatic mucinous cell hyperplasia (11)(12) and chronic pancreatitis (13)(14), a finding that may limit its value in pancreatic cancer diagnosis.

In a previous retrospective study (15), our group showed that K-ras mutation analysis in paraffin-embedded FNA samples from pancreatic masses contributed to cytological diagnosis in a substantial proportion of cases, mainly when suspicious cells, healthy-appearing duct cells, or insufficient material was reported, without false-positives cases. However, the clinical utility of this approach should be evaluated in a prospective setting, obtaining the molecular diagnosis in real time before the clinical decision is made for each individual patient. Moreover, no other studies have attempted to compare mutation analysis techniques with different detection limits. In the present study, we analyzed prospectively the diagnostic utility of K-ras mutation detection in 62 pancreatic FNA snap-frozen samples, using two artificial PCR-based techniques [standard and enriched restriction fragment length polymorphism PCR (RFLP/PCR)] that have distinct detection limits. Here we show that K-ras mutation analysis, prospectively performed in frozen FNAs, corroborates the previous retrospective findings in paraffin-embedded FNAs. Furthermore, the use of the enriched technique, which offers a lower detection limit, provides a better diagnostic sensitivity when compared with the standard technique evaluated previously (15). The enriched technique could be useful in the clinical setting, particularly in those institutions where the yield of cytology is moderate to low in the diagnosis of pancreatic cancer.

Materials and Methods

patients and samples

Between January 1995 and May 1997, 62 consecutive patients (32 men and 30 women, mean age of 63 ± 10 years) with pancreatic masses were included. In all cases FNAs of the masses were percutaneously obtained under ultrasonography (90%) or computed tomography (10%) guidance. A portion (50%) of each FNA was immediately examined as a fresh smear. The other 50% of the same cell suspension was snap frozen and stored at −80 °C. No attempt was made to examine the tumor cell content of the frozen samples. The procedures were in accordance with the standards of our institution’s ethics committee.

Final diagnosis of pancreatic carcinoma was established if malignant cells were identified in the FNA or in surgically resected specimens and/or when death occurred within the first year after diagnosis, with clinical evolution compatible with disseminated cancer disease. Other types of neoplasia were diagnosed on the basis of pathological findings. The diagnosis of chronic pancreatitis was based on standard clinical criteria and endoscopic retrograde cholangiopancreatography findings. In this set of patients, a minimum 6-month (range, 6–31 months) follow-up period with no evidence of cancer was available. Pancreatic tuberculosis was confirmed by positive Lowenstein culture. Final diagnoses were as follows: 46 pancreatic carcinomas, 2 mucinous cystic tumors, 5 other malignancies (1 lymphoma, 2 cholangio-carcinomas, and 2 lung metastases), 4 endocrine tumors, and 5 nonneoplastic diseases (4 chronic pancreatitis and 1 tuberculosis).

detection of K-ras codon 12 mutations

Enriched

BstNI RFLP/PCR method. DNA was extracted following standard procedures. We utilized a method that enriches for the amplification of mutant codon 12 K-ras alleles by cleaving amplified wild-type allele through intermediate digestion between first- and second-round PCR essentially as described by Kahn et al. (16). To create the restriction site for the enzyme BstNI [CCTGG], which is lost when a K-ras codon 12 mutation exists, the first-round amplification was performed using the mutant primers K-ras 5′ (17) and DD5P (sequences and PCR conditions shown in Table 1) in a volume of 50 μL containing PCR buffer (50 mmol/L KCl, 20 mmol/L Tris HCl, pH 8.4), 1.5 mmol/L MgCl₂, 0.2 mmol/L each dNTP (Promega Corp.), 1 U of Taq polymerase (Life Technologies Inc.), and 150 ng of PCR primers. The PCR reaction was performed in a Omnigene thermal cycler. An aliquot of 5 μL of the amplified product was enzymatically digested with BstNI following the manufacturer’s directions. One microliter of the digested product was reamplified using a heminested reaction with mutant amplimers K-ras 5′ and K-ras 3′ (30–35 cycles) (17). The latter primer artificially introduces an internal control to assure the completion of enzymatic digestion. After polyacrylamide gel electrophoresis (6%) and ethidium bromide (0.5 g/L) staining, the 143-bp band depicted the mutant allele, and the 114-bp band the wild-type allele. The positive control was NP9, a human pancreatic carcinoma cell line homozygous for an aspartic acid substitution at codon 12 of the K-ras gene. The negative control was NP18, a human pancreatic carcinoma cell line negative for the mutation.

To study the detection limit of the technique, serial dilutions of the mutant and wild-type alleles were prepared at the ratios shown in Fig. 1 . Criteria to deliver a positive result were established as follows. The RFLP/PCR approach used contains an internal control of enzymatic digestion to ensure its completion. A higher intensity of the 143-bp band, which depicts the mutant allele, when compared with the intensity of the 128-bp band, which depicts the internal control for enzymatic digestion, allowed identification of the mutant allele (Fig. 1) with confidence. When methods were standardized, a ratio of densitometric values of the 143- and 128-bp bands after ethidium bromide staining was obtained for controls and serial dilutions of wild-type and mutant alleles. No major densitometric signal of the 128-bp band was evidenced in the positive control. Ratios for serial dilutions were as follows: 1/10 = 10.2, 1/100 = 5.2, 1/1000 = 4.4, 1/10 000 = 2.2, negative control = 1.3. Naked-eye inspection by two independent observers (J.M. and P.P.) indicated differences in intensity when ratio values were equal and higher than 4.4. Consequently, this ratio should be considered as the cutoff for the presence of a mutant allele in the samples analyzed. When this criteria was used, the enriched RFLP/PCR method consistently detected a mutant allele in serial dilutions containing at least 1 mutant allele in 1000 wild-type alleles. Consequently, throughout the period studied we processed in every assay the control prepared at the detection limit of 10⁻³. Positive and negative controls for the mutation and controls for carryover DNA contamination, as well as the control prepared at the detection limit, were included in every experiment. Positive bands were always clearly identifiable when DNA obtained from FNAs was examined. Doubts concerning the positivity of the molecular diagnosis could not be raised by independent observers in any single case in our study. Moreover, all samples were analyzed in duplicate. Results were available 48–72 h after the tissue sample was obtained, and no radioactive material was needed. Clinicians were not informed of the result because this highly sensitive detection technique was under evaluation in our institution.

Standard RFLP/PCR method

Mutations at codon 12 of the K-ras gene were detected by means of the artificial RFLP/PCR method using two restriction enzymes: BstNI (17) (New England Biolabs Inc.) and HphI (18) (Amersham Life Science, Inc.).

The BstNI approach used was essentially the same as described in the enriched BstNI RFLP/PCR approach, with the only difference being that no enzymatic digestion was performed after the second-round amplification.

The HphI approach was used as described elsewhere (18). First-round amplification of exon 1 of the K-ras gene was done using K1USO and K1DSO primers and PCR conditions shown in Table 1 . The PCR reaction was performed under the same conditions as described above. One microliter of the amplified product was reamplified by means of a nested PCR using mutant amplimers K12U and K12D (Table 1). The K12D primer creates a restriction site for the enzyme HphI [GGTGA] (changing the second base of codon 13 G→A), which is lost whenever a mutation occurs in one of the first two bases of codon 12. The K12U primer artificially introduces an internal control to assure the completion of enzymatic digestion.

With the standard BstNI approach 1 mutant allele was detected when present in up to 100 wild-type alleles (Fig. 1). Similar results were obtained with the HphI approach. On the basis of results from our previous retrospective study using paraffin-embedded samples (15), these results were reported prospectively to the clinicians in charge of the patients.

Finally characterization of the detected mutations was performed by the single-strand conformational polymorphism (SSCP) method with silver staining as described (19).

Results

cytological examination

The presence of malignant cells was reported in 28 of 46 pancreatic carcinomas with no false positives. However, in 14 of the 46 FNAs of pancreatic carcinomas and in 1 FNA of chronic pancreatitis, the cytological report was not conclusive: 10 because of insufficient material and 5 because of the presence of suspicious cells (Table 2). In addition, healthy duct cells were found in four cases of pancreatic ductal adenocarcinoma (Table 2). The sensitivity of cytology was 75% in the diagnosis of pancreatic cancer (Table 3).

molecular diagnosis

Enriched

BstNI RFLP/PCR method. Molecular analysis was possible in 59 of 62 FNAs; in the remaining 3 cases amplification failed. K-ras mutations were detected in 35 of the 46 FNAs of pancreatic carcinomas, with no false positives (Table 2). The combination of cytology and enriched RFLP/PCR analysis was always informative and showed a sensitivity of 91%, with a specificity of 100% (Table 3). Only four pancreatic carcinomas failed to be correctly classified after the combined cytological and molecular analysis; one contained healthy-appearing duct cells, and three were reported as insufficient material (Table 2). When the enriched method was used, detection of K-ras gene mutations would have contributed to cytological diagnosis in 14 cases (Table 2): 6 of 9 with insufficient material, 5 of 5 containing suspicious cells, and 3 of 4 reported to contain healthy-appearing duct cells.

Standard RFLP/PCR method

Molecular analysis using the standard method was always feasible; K-ras mutations were detected in 29 of the 46 FNAs of pancreatic carcinomas with no false positives (Table 2). The combined cytological and molecular approach was always informative and showed a sensitivity of 80%, with a specificity of 100% (Table 3). The detection of mutations in the K-ras gene contributed to the diagnosis in nine cases of pancreatic cancer: four of five containing suspicious cells, three of nine with insufficient material, and two of four with healthy-appearing duct cells (Table 2). In our retrospective study (15), the absence of false-positive K-ras detection in FNAs of pancreatic masses strongly suggested malignancy when a K-ras mutation was found in the face of suspicious or insufficient material or healthy-appearing duct cells.

On the basis of these findings, molecular analysis did modify the clinical decision process in seven of the nine patients who were K-ras-positive and had no evidence of malignant cells, avoiding iterative fine-needle aspiration or further diagnostic procedures. In four cases, a K-ras positive analysis in combination with the presence of suspicious cells was considered confirmation of pancreatic cancer, and no further studies were performed. Two of these patients, in whom diagnostic laparoscopy was avoided, died 1 and 3 months later, respectively, with a clinical course consistent with advanced pancreatic cancer. In the other two, iterative fine-needle aspiration was avoided, and at surgery, positive peritoneal nodules or liver metastasis were present. In one patient with insufficient material at cytology, molecular analysis was the endpoint of the diagnostic work-up, and laparotomy was not performed because of the poor clinical status of the patient. Finally, in the remaining two patients with insufficient material for cytological evaluation and K-ras positive analysis, surgical resection of a histologically confirmed pancreatic carcinoma was performed. As we stated before, the detection of K-ras mutations did not modify the diagnostic schedule in two patients with healthy-appearing duct cells because the presence of malignant cells in the FNAs of concomitant hepatic nodules was diagnostic for disseminated disease.

characterization of K-ras codon 12 mutations by SSCP method

SSCP analysis was used exclusively for further characterization and confirmation of mutations detected by RFLP/PCR methods. No SSCP analysis was performed in the negative cases. Four of 35 FNA samples with positive K-ras mutations detected by the enriched BstNI method could not be characterized because of the higher detection limit of the SSCP method. Characterization of mutations was obtained after sequencing of PCR products (15) displaying abnormal mobility patterns and was as follows: 12 GTT (39%), 9 GAT (29%), 6 CGT (19%), and 4 TGT (13%).

Discussion

In the present prospective study, we compared two RFLP/PCR techniques for the detection of K-ras mutations, with different detection limits, for use in evaluating pancreatic masses. Both techniques contribute to the diagnosis when the cytological report is not conclusive. Although K-ras mutations have been detected previously in lesions of unknown malignant potential, mainly referred to as mucinous pancreatic duct hyperplasia frequently associated to chronic pancreatitis (11)(12)(13)(14), we have had no false positives in the evaluation of FNAs. Altogether, our results strongly support the use of K-ras mutation analysis with the enriched higher detection technique (1 mutant allele in the presence of up to 1000 wild-type alleles) in the diagnosis of pancreatic masses, which offers the best diagnostic sensitivity (91%) without false positives when combined with cytology. The main limitation of cytological analysis is the substantial proportion of cases in which a conclusive report is not possible, and where a second procedure to confirm diagnosis is required. Cytology does not detect malignant cells present in small proportions, either in the presence of healthy-appearing duct cells or when insufficient material is reported. The molecular approach allows detection of K-ras mutants even when cells are present in a small proportion. Mutation detection using the enriched technique would have complemented the cytological evaluation of FNAs in 14 cases of pancreatic cancer when cellular material was insufficient, suspicious cells were present, or when healthy-appearing duct cells were reported.

Only a minority of pancreatic cancers (four cases) failed to be correctly classified by the combined cytological and high detection molecular approach. Although inaccurate sampling of the lesion may account for some of the false negatives observed, the molecular approach has some limitations. In three cases, amplification failure occurred, probably because of low DNA concentration. Moreover, the low prevalence of K-ras mutations in pancreatic cancer in our population limits its usefulness; it is likely that the diagnostic sensitivity would improve in geographic areas with higher incidences of K-ras mutations (i.e., Japan or the United States) (6). Moreover, the intratumor heterogeneity for ras mutations already described may account for some false-negative results in the presence of accurate sampling (15). Finally, it is unlikely that the use of techniques for detection of codon 13 and 61 mutations would improve the present results because of the very low incidence of both genetic aberrations in human pancreatic cancer (6).

The detection of K-ras mutations in mucinous cell hyperplasia, evidenced in resected pancreatic masses developing in patients with chronic pancreatitis (11)(12)(13)(14), raised doubts about the specificity of mutation detection in the clinical setting. In the present prospective study, as well as in the previous retrospective (15), no mutations were detected in the FNA samples obtained from nonpancreatic carcinoma.

In the previous retrospective study (15), we used paraffin-embedded cell blocks to show that the standard technique (HphI and BstNI approach) could be of diagnostic utility in the evaluation of FNAs. The present study confirms the previous findings in a prospective design, obtaining the molecular diagnosis in real time, and shows that the molecular diagnosis can actually modify clinical decisions. Moreover, a positive molecular diagnosis avoided iterative pancreatic fine-needle aspiration or further diagnostic procedures in these patients. Unfortunately, no cost-benefit data could be obtained from our study because of the limited number of patients. Finally, although a cost-effectiveness study has not been performed, we are currently limiting the molecular analysis to those cases were the cytological report was not conclusive.

The clinical usefulness of ras mutations relies on the development of rapid, sensitive, and reproducible techniques for their detection. Several methodologies [i.e., allele-specific oligonucleotide hybridization alone or in combination with mutant-enriched PCR (20), ribonuclease A mismatch cleavage, artificial RFLP/PCR, SSCP, direct sequencing of the PCR product (8), or allele-specific amplification (21)] have been described. In the present study, detection of ras mutations at codon 12 was based on the creation of artificial RFLP using mutated primers (16)(17)(18) with some modifications. The method used herein offers a quick diagnosis (within 48–72 h), does not need radioactive material, and offers an excellent PCR yield. In a previous report by Urban et al. (22), PCR reactions were not successful in a large proportion (20%) of FNAs using fresh-frozen samples. The use of a two-round PCR technique may account for the increased PCR yield in our study. Molecular analyses in the clinical setting requires stringent controls, especially if two-round PCRs are used. Controls for carryover DNA contamination and controls prepared at the detection limit in each experiment validate the results. Moreover, the introduction of an internal control for completion of enzymatic digestion and the inclusion of positive and negative controls are mandatory to deliver a dependable result.

This study was supported, in part, by grants from Marató TV3 26/95, Agència d’Avaluació de Tecnologia Mèdica, and Comissionat per a Universitats i Recerca, Generalitat de Catalunya.

References