Pulmonary Squamous Cell Carcinoma following Head and Neck Squamous Cell Carcinoma: Metastasis or Second Primary?

Patients treated for head and neck squamous cell carcinoma frequently develop second primary tumors in the lung (1–3). Because the histologic appearance of the head and neck tumor and the lung lesion is similar (squamous cell carcinoma in both cases), the differential diagnosis between second primary or metastasis is mainly dependent on clinical criteria such as localization of the lung lesion, tumor stage, and disease-free interval. Warren and Gates (4) were the first to develop criteria for second primary tumors and additional studies have been published since (summarized in Table 1). The criteria employed, such as histology, tumor stage, and interval, however, do not give a definitive answer in the differential diagnosis mentioned above.

Treatment options for patients with second primary lung tumors are different from those for metastasized head and neck squamous cell carcinoma. A resectable second primary would justify surgical approach with curative intent (5–9), whereas metastatic disease would be treated with systemic therapy or limited surgical resection. Therefore, we looked for the identification of more objective markers in this differential diagnosis.

In theory, clonal relationship between two tumors proves their common origin in case of metastases. Conversely, the absence of a clonal relationship would suggest a second primary carcinoma. Many assays have been developed to evaluate clonal relationship between tumors. Analysis of patterns of loss of heterozygosity (LOH) has been used to distinguish second primaries from metastases in the aerodigestive tract (10–12). LOH analysis is easy to perform on paraffin-embedded tissue and, thus, potentially very useful in everyday clinical practice, as well as for the analysis of retrospective series. In this study, we have compared LOH patterns in head and neck squamous cell carcinomas and squamous cell carcinomas of the lung to evaluate their clonal relationship. These results were compared with the results using clinical and histologic criteria. We discuss whether LOH analysis is an adjunct to these conventional criteria and how it can be used on a routine basis for clinical decision-making.

Patient. All cases between 1978 and 2002 with head and neck squamous cell carcinoma and malignant squamous lesion of the lung were included provided that sufficient histologic material was available at the Department of Pathology of the Netherlands Cancer Institute. In 44 patients, these criteria were met. Of these 44 patients, 36 were males and 8 were females. The average age at metastasis was 62 years (range, 26-87 years). Twenty-two patients had laryngeal carcinoma: 10 oral cavity carcinoma, 8 oropharyngeal carcinoma, and 4 hypopharyngeal carcinoma.

These cases were found using the PALGA database (Pathologisch Anatomisch Landelijk Geautomatiseerd Archief, i.e., pathologic anatomic national automated archive of the Netherlands).

Clinical criteria. The clinical parameters were recorded, as follows:

1. Characteristics of the head and neck tumor

   • Stage of head and neck tumor at presentation according to IUCC guidelines 2002

   • Locoregional recurrence at time of pulmonary lesion

2. Presentation of the lung lesion

   • Location of the lung lesion: central or peripheral

   • Number of lung lesions: solitary or multiple

3. Time interval between first admission for the head and neck squamous cell carcinoma and radiological detection of the pulmonary lesion

If two of the three items mentioned above were in favor of metastasis or second primary, a lung lesion was clinically classified as such. Thus:

• Ad1. Stage I or stage II and no signs of local or regional recurrence were in favor of second primary lung tumor. Stage III or IV or signs of locoregional recurrence were in favor of metastasis.

• Ad2. Central location and solitary lesion were in favor of second primary tumor. Peripheral location or multiple lesions were scored in favor of metastasis.

• Ad3. If the interval between the first admission for the head and neck squamous cell carcinoma and the radiological detection of the lung lesion was longer than 3 years, it was scored in favor of a second primary lung tumor; if shorter, this scored in favor of metastasis.

Histology. Routine H&E sections were used for tumor classification according to WHO criteria (13). Only in case of a specific squamous cell carcinoma subtype (for instance basaloid squamous cell carcinoma) histology was considered conclusive in the differential diagnosis of second primary versus metastasis. Subsequently, the H&E slides were used to identify an area within the block with more than 50% tumor cells. This area was microdissected for DNA isolation.

Loss of heterozygosity detection. Serial 10 μm paraffin sections were deparaffinized by standard procedures and incubated for 16 hours in 1 mol/L sodium thiocyanate at 37°C to remove cross-links, followed by 2 × 5-minute wash in PBS (Dulbecco). Subsequently, guided by the adjacent thin H&E section, the region containing over 50% tumor cells was microdissected from the glass slide using a scalpel. The tissue was then transferred to a tube containing digestion buffer: 2 mg/mL proteinase K (Roche, Basel, Switzerland) in 10 mmol/L Tris-HCl (pH 8.0), 50 mmol/L KCl, 2.5 mmol/L MgCl₂, 0.5% Tween 80, 0.1 mg/mL gelatin. Tubes were incubated for 24 hours at 55°C; proteinase K was then heat inactivated at 95°C for 10 minutes, and after centrifugation, the supernatant was transferred to a clean tube and stored at 4°C until use.

DNA isolated from the tumor as well as normal (not tumor) tissue obtained from a separate tissue block was analyzed using 12 of the following markers (D1S158, D1S228, D1S2766, D3S1038, D3S1270, D3S158, D4S403, D4S430, D4S1572, D4S3046, D5S346, D5S644, D6S1719, D8S552, D9S63, D9S288, D9S286, D10S217, D11S4090, D11S4190, D13S158, D14S65, D17S799, D18S58, D18S474) distributed over 11 chromosome arms using PCR. These markers were selected within chromosomal regions that show frequent loss in lung or head and neck cancer according to the literature (14, 15). The fluorescent label of the primer allows the semiquantitative evaluation of the PCR products using the ABI 3700 capillary DNA sequencer (Perkin-Elmer Biosystems, Branchburg, NJ). The peak intensity of the PCR products was measured and used to calculate the intensity ratio between the two allele peaks of heterozygotes. This ratio was subsequently divided by the ratio of the normal alleles for normalization resulting in a LOH index. A LOH index below 0.75 was interpreted as a LOH; above this value, it was considered as retention of heterozygosity (ROH; ref. 16).

Analysis of loss of heterozygosity pattern. Calculation of the probability of occurrence of concordant LOH patterns by chance (instead of due to clonal relation) was not allowed as the occurrence of LOH on different chromosome arms was not an independent event (see Results) as is also known from literature (17). Because there is also no consensus procedure to evaluate the use of LOH patterns in deciding on clonality, we set out to develop a new strategy. The LOH patterns of the first 22 tumor pairs were interpreted by three experts (P.N., L.v.t.V., and D.d.J.). Three basic principles were then formulated, as follows:

1. Discordant LOH (e.g., LOH of different alleles) not explainable by progression (LOHx-LOHy or LOH-ROH) is a strong argument for second primary tumor as this pattern is virtually impossible when two tumors are clonally related and, thus metastasis.

2. Concordant LOH [e.g., LOH of the same allele (LOHx-LOHx)] is an argument for clonal relation, but can also occur by chance especially if the expected frequency of LOH is high.

3. ROH-LOH is a variant of discordant LOH that could occur as a result of tumor progression and is therefore a less strong argument against clonal relation (metastasis) than the two other forms of discordant LOH (LOHx-LOHy and LOH-ROH).

These principles were formalized in a diagram (Fig. 1). According to this diagram, discordant LOH not explainable by progression (LOHx-LOHy or LOH-ROH) is considered first. If present for two or more alleles on two or more different chromosome arms, the second tumor is considered a second primary tumor. If this is not the case, the number of chromosome arms with concordant LOH (LOHx-LOHx) is considered in the second step. If this is present on four or more chromosome arms, the second tumor is considered a metastasis. If concordant LOH is present for two or three alleles on two or three chromosome arms, the second tumor is considered a probable metastasis. If concordant LOH is not present or only for one allele, all forms of discordant LOH, thus including ROH-LOH (which might be explained by tumor progression), are considered in a third step. The 45 tumor pairs were then categorized according to this diagram. These results were used for further analysis.

General. The institutional review board has reviewed the completed manuscript. All specimens were handled in a coded fashion according to the ethical guidelines of the Dutch federation of medical science (FEDERA).

Clinical scoring

1. Characteristics of the head and neck tumor

   • Stage I, II, III, and IV primary lesions were seen in 5, 9, 12 and 18 patients, respectively (Table 2, “Stage”). Corresponding tumor and node stages are summarized in Table 3.

   • Fourteen patients had a local recurrence at time of the pulmonary lesions; in six of these patients, in spite of initial stage I or II disease (Table 2, “Recurrence”).

2. Presentation of the lung lesion

   • Seven patients presented with a central lung lesion; one of these patients had multiple lung lesions (Table 2, “Central, peripheral”).

   • Twenty-four patients presented with a solitary lung lesion. Eight patients presented with two lung lesions, whereas the other 12 patients had multiple lung lesions (Table 2, “Solitary, multiple”).

3. Time interval between first admission for the head and neck squamous cell carcinoma and radiological detection of the pulmonary lesion.

   • The interval between the first admission for the head and neck squamous cell carcinoma and the radiological detection of the lung lesion was longer than 3 years for 5 patients and shorter for the other 38; nine of them had an interval between 2 and 3 years. One patient developed head and neck squamous cell carcinoma after lung carcinoma and therefore was scored as second primary carcinoma. The detailed interval in months is shown in Table 2. Using these clinical criteria, six patients were considered to be likely to have a second primary lung tumor, whereas 38 patients were thought to have lung metastasis (Table 2, “Clinical conclusion”).

Using these clinical criteria, six patients were considered to be likely to have a second primary lung tumor, whereas 38 patients were thought to have lung metastasis (Table 2, “Clinical conclusion”).

Histology. Thirty-one tumor pairs were of a concordant histologic type: 22 tumor pairs were both keratinizing squamous cell carcinoma, 6 nonkeratinizing, 2 poorly differentiated, and 1 basaloid. Thirteen tumor pairs showed discordant histologic features. Of eight tumor pairs, one tumor was keratinizing whereas the others were not; in five tumor pairs, a keratinizing tumor and a poorly differentiated lesion were seen. One patient had several lung lesions of which one proved to be an adenocarcinoma whereas the others were conventional keratinizing squamous cell carcinoma. The adenocarcinoma was considered a second primary tumor on histologic grounds.

Loss of heterozygosity analysis. LOH results per marker per patient are shown in Fig. 2. Overall LOH frequency in the 449 informative markers was 224 (49.9%) in the head and neck cancers and 242 (53.9%) in the lung lesions. It varied from 0% to 100% between lesions in both head and neck (Fig. 3, solid line, shaded) and lung. This variation between patients is much larger than could be expected if LOH of different markers within a patient would be independent (χ²: P < 0.001). The broken line in Fig. 3 shows the expected distribution of LOH frequency if LOH for different markers within one tumor would have been an independent event. As the occurrence of LOH was not independent, calculation of probability of LOH patterns was not allowed. Therefore, as mentioned in the Materials and Methods, a diagram for interpretation of LOH patterns was developed (Fig. 1).

Applying this diagram, 11 tumor pairs were classified as metastasis and 8 as probable metastases. Twenty-two cases were classified as certain and three as probable second primary tumors. In one case, the LOH analysis was inconclusive. In this case, there was not enough concordant or discordant LOH to draw any conclusions.

Integration of clinical, histologic, and molecular biological results. All six patients that were clinically classified to have a second primary lung tumor (Table 2, “Clinical conclusion”), as well as the case that was histologically classified as a second primary, were classified as second primaries by LOH analysis as well. Four of these tumor pairs were by histology both keratinizing squamous cell carcinoma or nonkeratinizing squamous cell carcinoma. Thus, histology was the same although these tumor pairs were neither clinically nor molecular biologically metastasis.

Nineteen tumor pairs were classified as metastasis both on clinical grounds and by LOH analysis. Four of these tumor pairs showed variation of histologic appearance between the lung and the head and neck lesion. One of these pairs showed keratinizing head and neck lesion, whereas the lung lesion was undifferentiated. The other three pairs were keratinizing versus nonkeratinizing. Thus, using the proposed interpretation scheme, there are no discordant cases in those classified as metastasis by LOH analysis or in those clinically classified as second primary, but histology is of little help in this classification, being as often concordant as discordant in either related or unrelated tumor pairs.

Although LOH analysis supported the clinical scoring for 25 patients, it suggested a different origin for the lung lesion in 18 cases. These 18 discordant cases all were considered to be metastasis according to clinical characteristics, whereas by LOH analysis, a second primary was thought more likely. In one case, LOH analysis was inconclusive.

To investigate if the first assumption in our interpretation scheme for LOH analysis (Fig. 1) was right [e.g., LOHx-LOHy or LOH-ROH (discordant LOH not explainable by progression) is a strong argument in favor of second primary tumor], we reanalyzed the LOH results starting with the assumption on the second line of Fig. 1: “concordant LOH (LOHx-LOHx) of more than one allele is a strong argument in favor of clonal relation, i.e., metastasis.” Using this strategy, eight cases were classified differently (Table 2, MB2). Two of these were the clinically classified as second primary. One of them was the case histologically classified as second primary tumor (the adenocarcinoma). As a second alternative interpretation strategy, we augmented the threshold in the first line of Fig. 1 to three alleles with discordant LOH (not explainable by progression). This resulted in five cases to be classified differently as compared with the first classification scheme (Table 2, MB3 versus MB1). Two of these five cases were initially classified as second primary both clinically as well as by LOH analysis. Thus, our first interpretation strategy resulted in the best match with those cases clinically classified as second primary tumor.

As there is no gold standard in the differential diagnosis of metastasis versus second primary tumor in patients with a squamous cell lesion of the lung and a squamous cell carcinoma of the head and neck region, we compared the value of LOH analysis, histology, and clinical classification in this dilemma. LOH analysis showed that about half of the lung lesions (18 of 38) that would clinically be classified as metastasis were suggested to be second primary tumors. Histologic classification was of negligible help.

For clinical classification, all criteria frequently mentioned in literature (Table 1; refs. 2, 4–7, 11, 18–26) were used to categorize a lung lesion either as metastasis or as second primary tumor. This resulted in a conservative classification of second primary tumors (i.e., a lung lesion was only considered a second primary if the majority of signs were against metastasis). As a result, in this study, only 6 of 44 lung lesions were clinically classified as second primaries.

Histology did not contribute to the decision-making in all but two cases. All tumors were squamous cell carcinomas with a variable degree of keratinization, except one adenocarcinoma. In this case, the adenocarcinoma was one of several lung lesions but was classified as a second primary tumor on histologic grounds. The other case showed the uncommon basaloid variant of squamous cell carcinoma in both lesions and, therefore, histology was thought conclusive for metastasis. In both cases, the conclusion on histologic grounds was confirmed by LOH classification.

To show clonal relationship between two lesions, several molecular biological techniques are available. LOH analysis can now easily be done on paraffin-embedded material, and is therefore practical to perform in daily clinical practice. Other techniques, such as p53 mutation analysis or X chromosome inactivation, are either very time-consuming or only applicable in a restricted population (women). Therefore, we tried to classify the lesions with LOH analysis only.

In this study, we used microsatellite markers on 11 different chromosome arms, whereas others analyzed multiple markers on two to six chromosome arms (6, 10, 27, 28). In general, clonality was decided on the basis of concordant LOH of at least two loci, whereas in some articles the additional criterion of identical p53 mutations was used, if present (6, 10, 27, 28). This approach, however, overlooks the phenomenon that LOH on different loci is not an independent process as we and others (17) have shown. Therefore, we developed an interpretation scheme that integrated this knowledge of dependence as well as biological insight (Fig. 1). In this strategy, discordant LOH not explainable by progression (LOHx-LOHy or LOH-ROH) is used as the most important feature as it is unlikely to occur in clonally related tumors. ROH-LOH (loss of heterozygosity in the lung tumor but not in the earlier head and neck tumor) is less important as it can well be explained by tumor progression (accumulation of genetic changes within the tumor). Concordant LOH (LOHx-LOHx) is only considered in the second step as it can occur very easily by chance in unrelated tumors, given the high frequency of LOH as observed in this study (33-82%) as well as by others (10, 17, 27) and given the dependence described in this article. Classification using this scheme resulted in a 100% match with clinical scoring as far as the second primary tumors were concerned (6 of 6), whereas 18 of 38 lung lesions clinically classified as metastasis were considered second primary tumors (Table 2, MB1).

The results of alternative schemes were explored (Table 2, MB2 and MB3). Both schemes revealed cases that were clinically classified as second primary but using the alternative scheme would be classified as metastases. As discussed before, the clinical criteria for second primary tumor were very stringent. It is therefore unlikely that these cases are metastasis. Furthermore, these cases had discordant “LOH not explainable by progression” for at least two markers (Fig. 2, patients 10, 18, 19, 20, 22, and 23). Although discordant LOH in two related tumors can be explained by tumor heterogeneity (29), this needs several events in different parts of one tumor and is thus less probable than LOHx-LOHx by chance. In addition, the adenocarcinoma would have been classified as a metastasis using these schemes. Nonetheless, 12 or 16 of 38 cases clinically classified as metastasis would be classified as second primary tumors using these alternative schemes for LOH interpretation.

Thus, PCR-based analysis of LOH is a feasible and useful addition in daily clinical practice for the differential diagnosis of metastasis versus second primary tumor in case of lung lesions after treatment for primary head and neck squamous cell carcinoma. An interpretation scheme is developed with emphasis on discordant LOH results, whereas concordant LOH is only taken into account as a second step in interpretation of LOH patterns. Still, whatever classification scheme is used to interpret LOH results, many lung lesions clinically interpreted as metastasis are second primary tumors when classified after LOH analysis, justifying a surgical approach with curative intent in these cases.

We thank Prof. C.J. Cornelisse and Dr. L. Wessels for stimulating discussions.