Prognostic Value of Genetically Diagnosed Lymph Node Micrometastasis in Non-Small Cell Lung Carcinoma Cases¹

Numbers of operable NSCLC³cases are increasing because of improved detection of the disease with sophisticated diagnostic techniques. However, even after complete or potentially curative surgery, the long-time survival rate remains unsatisfactory (1). This is related to the fact that disseminated tumor cells, undetectable by routine diagnostic methods,may be present at the time of surgery. If they could be detected,prediction of relapse or prognosis would be facilitated, allowing additional appropriate treatment to be performed for high-risk patients. Micrometastasis to regional lymph nodes, not detected by routine histology, can now be identified by immunohistochemical or genetic methods, and this is reported to be useful for assessment of prognosis (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12). High false-positive rates are a major problem, but the MASA genetic method has high sensitivity and specificity for carcinomas; one tumor cell containing genetic changes was detected in a background of thousands of normal cells(13, 14, 15, 16, 17). It has established prognostic value for colon cancer (2, 13), but data for lung cancer are limited. One reason is that it is difficult to collect enough cases of early stage carcinomas having genetic alteration for MASA analysis. The other is that amplification of the mutated DNA sequence in lymph nodes presented practical difficulties in the past.

In the present study, we therefore collected many NSCLCs with mutations and analyzed regional lymph node micrometastasis by a modified improved MASA method and assessed its prognostic potential. For comparison, the immunohistochemical method using a monoclonal anti-CK reagent was also performed for the same specimens.

Fresh frozen samples from 151 NSCLCs (113 adenocarcinomas and 38 squamous cell carcinomas), which were resected consecutively from 1989 to 1993 at Cancer Institute Hospital, Tokyo, Japan, were screened for p53 mutations in exons 4–8 and 10 using single-strand conformation polymorphism, followed by sequencing, and for K-ras mutations of codons 12, 13, and 61 using the MASA method, as described previously (18, 19, 20). All patients analyzed had undergone a complete or potentially curative resection with lobectomy or pneumonectomy, combined with pulmonary hilar and mediastinal lymph node dissection. Histopathological classification and differentiation of the tumors were determined according to the 1981 WHO classification of lung tumors (21). The stage of the disease was based on the TNM staging system of the International Union against Cancer (UICC; Ref. 22). Of the 68 p53and 12 K-ras mutated cases, 35 cases with p53mutations and one case with a K-ras mutation were staged as pathological (p)-T_1–4 N₀and N₁ M₀. Of those cases,5 cases with p53 mutation were excluded because the PCR products with the MASA method for the positive controls could not be consistently detected. Finally, 31 cases staged as p-T_1–4 N₀ and N₁ M₀ with enough follow-up were eligible for analysis (Table 1). Location and type of mutations for each patient are shown in Table 2. None had received chemotherapy or radiotherapy before surgery, but 13(41.9%) underwent postoperative adjuvant therapy. The patients comprised 24 men and 7 women with a mean age at surgery of 60 years(range, 26–76 years). Twenty-two had adenocarcinomas, and 9 had squamous cell carcinomas, with 8 well, 16 moderate, and 7 poorly differentiated lesions. Two hundred and twenty-nine lymph nodes(median, 8; range, 4–9 for each case) from the 31 patients were eligible for comparison of the micrometastasis positivity between MASA and immunohistochemical methods and for evaluation of prognostic value(Table 3). One hundred and seventy lymph nodes of 22 p-N₀(no lymph node metastasis by conventional histological examination)cases were examined in both pulmonary hilar and mediastinal regions. For the remaining nine p-N₁ patients, 59 lymph nodes from mediastinal regions were examined by both methods. Survival duration was calculated from the date of operation until the date of the last follow-up (censored) or the date of death. The end point was cancer-related death, and deaths attributable to other causes were treated by censoring. Complete follow-up information was available on all 31 patients, with a median follow-up period of 73 months (range,8–108). Within 5 years after surgery, six patients died with distant or intrapulmonary metastasis, and the others, with the exception of 1 patient who died of another cause (case no. 8), were free of disease(Table 3).

Lymph nodes fixed in formalin and embedded in paraffin, without metastasis by conventional histological study, were analyzed for micrometastasis with serially sectioning. The first and last 4-μm sections were stained with H&E to confirm or refute the presence of overt lymph node metastasis by an experienced histopathologist(E. T.). Genomic DNA was extracted from the second and third 15-μm sections and used for genetic analysis (23). The remaining one 4-μm section was examined by IHC.

DNAs used for nested-PCR were extracted from histologically negative lymph nodes. As positive controls, DNAs were extracted from primary tumors fixed in formalin and embedded in paraffin for each patient, and diluted tumor DNAs (1:10, 1:100, and 1:1000) with DNAs extracted from lymph nodes of other cases were used. Also, lymph node DNAs extracted from other cases were used as negative controls. For nested PCR, two sets of primers for each patient were prepared. For the first-round PCR amplification, primers located outside of the mutational site were synthesized to amplify the products within 150 bp, because genomic DNAs from tissues fixed in formalin and embedded in paraffin were in fragments. Mutant allele-specific synthesized primers of 15–22 bp with 3′-ends corresponding to each p53 or K-rasvariant were used for the second round of PCR (MASA method; Table 2;Refs. 13 and 16).

Genomic DNAs from lymph nodes were subjected to first PCR amplification with the set of outer primers (0.5 μm) in 25-μl mixtures consisting of 10 mm Tris-HCl (pH 8.3), 50 mm KCl, 1.5 mm MgCl₂,0.01% gelatin, 0.2 mm deoxynucleotide triphosphates (dATP,dTTP, dGTP, and dCTP), and 0.625 unit of AmpliTaq GOLD (Perkin-Elmer Corp., Foster City, CA), with denaturing at 95°C for 9 min, annealing and extension at 52–66°C for 30 s in each cycle, using a GeneAmp PCR System 9700 (Perkin-Elmer Corp.). Because of varied qualities of genomic DNAs from lymph nodes, PCR were performed for 45–50 cycles to obtain equivalent amplified products by saturation. After purification for removal of the primers and deoxynucleotide triphosphates, the amplified products were diluted 50-fold in 10 mm Tris (pH 8)-1 mm EDTA buffer, and 1 μl of aliquots of diluted products was subjected to a second round of PCR using nested primers (mutant allele-specific primers) under the same conditions, except for the temperature of annealing and extension(64–76°C) and the figures of PCR cycles (35–41 cycles), which differed with each set of primers. Primer sequences and PCR conditions of each primer set are shown in Table 2. The PCR products were then subjected to 3% agarose gel electrophoresis and visualized by ethidium bromide staining. All specimens were analyzed at least twice to confirm the results. Furthermore, in the first two cases analyzed, we examined the sequences of the final MASA products and confirmed their identity as those targeted.

Paraffin-embedded tissue sections 4 μm thick were immunostained with the monoclonal anti-CK reagent, CAM5.2 (Becton Dickinson, San Jose,CA), using an immunoperoxidase method. This primary mouse monoclonal antibody is specific for CKs 8 and 18 of epithelial cells and is positive in both adenocarcinomas and squamous cell carcinomas of the lung (24). After exhaustion of endogenous peroxidase with hydrogen peroxide, deparaffinized sections were incubated with 0.1%trypsin for 45 min at room temperature to expose antigen sites. Possible nonspecific background staining was blocked with 10% normal goat serum for 10 min at room temperature. Then the sections were incubated with a primary antibody at room temperature for 1 h, and the antibody reaction was developed with secondary antibodies using ENVISION polymer reagent (peroxidase-labeled polymers conjugated to goat antimouse and antirabbit immunoglobulins; Dako, Glostrup,Denmark). Next, 0.02% diaminobenzidine was applied as the chromogen,and the sections were counterstained with hematoxylin. For positive controls, tissue sections of both primary tumors and metastatic lymph nodes detected by conventional histopathology were included in each staining batch. Negative control sections for IHC were stained with the primary antibodies omitted.

To assess any correlations between the presence of micrometastasis and clinicopathological data, Fisher’s exact probability test,Mann-Whitney’ U test, and Student’s t test were used, with P < 0.05 indicating significance. Survival curves were created by the Kaplan-Meier method, and the statistical significance of differences was calculated by the log rank test.

Histologically, no lymph node metastases were reconfirmed, even with retrospective examination of the H&E slides after obtaining genetic and IHC results.

Representative results of the MASA method are shown in Fig. 1, and a summary of the results is presented in Tables 1 and 3. The genetic method revealed 34 (14.8%) of 229 lymph nodes and 9 (29.0%)of 31 cases to be positive. Among 22 patients with p-T_1–4 N₀M₀ disease, 24 (14.1%) MASA-positive lymph nodes of a total of 170 in the hilar and/or mediastinal regions were detected in 6 (27.3%) patients; micrometastases in two patients were limited to pulmonary hilar lymph nodes and 4 to mediastinal regions. Among nine cases with p-T_1–4 N₁M₀ disease, 10 (16.9%) of 59 lymph nodes in the mediastinal region were MASA positive in 3 (33.3%) patients.

CK-positive cells were clearly stained dark brown in their cytoplasm and cell membranes. All primary tumors were positive for staining with anti-CK regent, strongly in all adenocarcinomas but weakly in some cases of squamous cell carcinomas. Lymph nodes were considered positive even if only one CK-positive cell was detected within a lymph node. An example of a CAM5.2-positive cell by the immunohistochemical method is shown in Fig. 2. The immunohistochemical approach demonstrated positive reactions in 61(26.6%) of 229 lymph nodes (24 cases, 77.4%) among patients with p-N₀ and p-N₁: 47 (27.6%)of 170 lymph nodes (17 cases, 77.3%) with p-N₀,and 14 (23.7%) of 59 lymph nodes (7 cases, 77.8%) with p-N₁ patients (Table 3). Among 17 p-N₀ patients with CK-positive reactions,positive lymph nodes were limited to the pulmonary hilar region in eight cases and up to the mediastinal region in nine cases. The concordance rate between the two methods was 76.9%, but 13 (39%) of 34 genetically positive lymph nodes were not positive with IHC.

Among 22 p-N₀ patients, 4 (66.7%) of 6 with micrometastasis detected by the genetic method and none of those without micrometastasis died by cancer relapse. Fig. 3 A shows the survival curves of the p-N₀ patients with and without micrometastasis (P < 0.001,log rank analysis). Among the same p-N₀ patients,only 4 (23.5%) of 17 CK-positive cases died by relapse, and although the remaining 13 (76.5%) patients were censored, 5 CK-negative patients did not relapse. No prognostic difference was observed between patients with and without micrometastasis decided by IHC(P = 0.246).

Fig. 3,B shows survival curves according to the N factor assessed by conventional histology. Survival duration between p-N₀ and p-N₁ patients did not differ (P = 0.856). However, when the N factor was restaged according to the genetic results (g-N), patients with g-N₁ and g-N₂ disease showed a shorter survival than those with g-N₀(P = 0.042 and P < 0.001, respectively; Fig. 3,C). The 5-year survivals were 100.0% for patients with g-N₀, 75.0% for g-N₁, and 42.9% for g-N₂. When restaged by IHC (IHC-N), survival between patients with IHC-N₀ and IHC-N₁ or IHC-N₂ did not show any statistically significant difference (P = 0.289 or P = 0.239, respectively; Fig. 3 D).

In the present study, a conventional MASA method was initially applied to analyze mutations in DNAs extracted from formalin-fixed,paraffin-embedded lymph nodes, but the results were sometimes equivocal. Because the main reason was probably that DNAs extracted from these samples were fragmented and not uniform, we first amplified those DNA fragments that include the mutated region to unify the quality of samples, and then MASA analysis was performed using amplified DNA, in which targeted sequences could be reliably detected. Furthermore, we used several positive and negative controls for confirmation. As positive controls, primary tumor DNAs and diluted DNAs were used. The dilution fold was determined on the basis of initial experiments for several cases in which diluted tumor DNAs at 10-, 100-,1,000-, and 10,000-fold were amplified with the modified MASA method. The results showed that in all cases a MASA-specific product was constantly observed in the range from 10- to 1,000-fold but not always identified at 10,000-fold. By this method, we demonstrated that ∼30%of patients with histologically negative lymph nodes indeed had micrometastases, and that these were linked to a poor outcome.

Thus far, for detection of micrometastasis in lymph nodes the immunohistochemical method with anti-CK regent has been used frequently and reported to be effective for prediction of a poor prognosis(3, 4, 5, 7, 8, 9). However, our results showed the modified MASA method to be superior to IHC in terms of NSCLC relapse. In regional lymph nodes, macrophages that phagocytize CKs derived from degraded normal lung epithelial cells are presumably present, and these could be misinterpreted and give rise to false-positives. In fact, we could not recognize carcinoma cells histologically in CK-positive lymph nodes. Furthermore, there have been reports that CK expression may be detected in both normal and neoplastic nonepithelial cells (11, 25, 26).

As a genetic approach to detect lymph node micrometastasis or systematic tumor dissemination, reverse transcription-PCR analysis for several markers, including CK19, has been described (10, 11, 12, 27). Although having a high sensitivity, this is again plagued with the problem of a high percentage of false-positive results,depending on the number of PCR cycles, because of nonspecificity for cancer cells (11, 12, 27). The modified MASA method applied here is not only highly sensitive but also highly specific for carcinoma cells with particular mutations, so that we could show clearly the correlation between the micrometastasis and postoperative survival in lung carcinoma patients. As to mutated DNAs, they may be present in macrophages if they phagocytize carcinoma cells or carcinoma DNAs (28), but this would reflect the likelihood of metastasis.

Carcinoma relapse was found in contralateral lungs or other remote organs in our patients with micrometastasis. Two pathways of metastasis are probable; one is that carcinoma cells pass through the regional lymph nodes and reach blood vessels, and the other is that vascular invasion of carcinoma cells occurs at the primary site. In our study,all six patients with relapse had lymph node metastasis histologically and/or genetically, and in four patients of them metastasis extended to mediastinal region. However, none of the patients without such lymph node involvement showed recurrence. Therefore, the main root for systemic metastasis may be via lymphatic pathways, which strengthens the importance of micrometastasis in lymph nodes. However, in two relapsed patients, lymph node metastasis was limited to the hilar region, so that the second route cannot be ruled out.

If lymph node or distant metastasis is within a “micro level,”undetectable by routine histological examination, intensive adjuvant therapy may prevent carcinoma recurrence. Therefore, a finding of micrometastasis in lymph nodes is important not only for prediction of prognosis but also for selection of patients for whom postoperative extensive adjuvant chemotherapy may be performed to advantage. Further analysis of a large number of patients with this method, possibly in a prospective fashion, is warranted to confirm our results.

We thank Drs. H. Sugano (Cancer Institute), Y. Nakamura(Institute of Medical Science, University of Tokyo), S. Sugai (Cancer Institute), and T. Kozu (Saitama Cancer Center Research Institute) for helpful advice and discussions. The technical assistance of T. Yoshikawa and Y. Yamaoka is gratefully acknowledged.