Novel Approaches in the Treatment of Non-Small-Cell Lung Cancer

Introduction

To date, the treatment of non-small-cell lung cancer (NSCLC) has been entirely empirically based. That is, if each drug has a 20% response rate, then the combination of drugs A and B easily predicts a 40% response rate and will be rapidly accepted if both drugs have different, nonoverlapping toxicity profiles. A plethora of new cytotoxic anticancer agents have increased response rates—either as a single drug or in combination. However, the results of a recent, four-arm randomized Eastern Cooperative Oncology Group study[1] indicate that no particular combination of either cisplatin(Drug information on cisplatin) (Platinol) with paclitaxel(Drug information on paclitaxel) (Taxol), docetaxel(Drug information on docetaxel) (Taxotere), or gemcitabine(Drug information on gemcitabine) (Gemzar) surface as a clear salient chemotherapy combination. Certainly new approaches need to be implemented in the treatment of non-small-cell lung cancer.

Rational Background: Genetic Markers In Colorectal Cancer

The Human Genome Project has almost been completed, and a vast majority of approximately 10⁵ human mRNA have been at least partly cloned and sequenced as cDNAs. When completed, these cDNAs will provide valuable tools to study the protein products of their respective mRNAs. These diverse sets of proteins could be selected for anticancer drug discovery programs.[2] New knowledge of normal and aberrant function of oncogenes and tumor suppressor genes provides unique opportunities to understand, and ultimately to control, the processes leading to malignancy. Thus, identification of the molecular and genetic events involved in each step of tumor progression may be central to understand NSCLC and to develop resistant markers and novel targets for pharmacologic intervention in this tumor.

Using laser capture microdissection and cDNA arrays, it has been found that most squamous cell carcinomas of the head and neck overexpress members of the wnt and notch growth and differentiation regulatory system. This suggests that the wnt and notch pathways may contribute to squamous cell carcinogenesis.[3] Also, high levels of expression of notch 3 (a member of the notch proto-oncogene family) have been observed in 25% of NSCLC patients tested. Overexpression was correlated with translocations, involving 19p in these lines.[4]

Gemcitabine Trial

A randomized trial showed that gemcitabine as a single agent yields the same response rate and survival as the cisplatin/etoposide combination but with much better tolerability. The response rate was 17.9% for the gemcitabine arm and 15.3% for the cisplatin/etoposide arm. Median survival times were 6.6 months for gemcitabine and 7.6 months for cisplatin/etoposide. The 1-year survival probability estimate was 26% for gemcitabine and 24% for cisplatin/etoposide. Also, patients in the gemcitabine arm had a greater probability of achieving a tumor response after 2 months (8% vs 0%) and lasting at least 6 months (73% vs 45%).[5]

Docetaxel, Paclitaxel

Treatment of non-small-cell lung cancer, however, remains an almost insurmountable pitfall for oncologists. In randomized studies,[6,7] docetaxel and paclitaxel produced major objective tumor regression in only 13% to 16%. But these single agents still induce significant survival improvement over best supportive care. In the paclitaxel randomized study, median time to progression was 4 months with paclitaxel and only 1.2 months for best supportive care alone (P = .001).[7] Recently, on the grounds of meta-analysis of individual data from a large colorectal cancer clinical trials dataset, Buyse et al[8] proposed that response rate could be a surrogate end point for survival advantage.

Irinotecan, Oxaliplatin(Drug information on oxaliplatin)

Clinical trials in metastatic colorectal cancer have shown that any improvement in survival is modest albeit statistically significant, even with regimens containing new agents such as irinotecan(Drug information on irinotecan) (CPT-11, Camptosar,) or oxaliplatin (Eloxatine). Striking improvements in survival are unlikely until the complete response rate is high.[9] Interestingly, pharmacogenetically oriented treatment has started for colorectal tumors where response and survival have been associated with the thymidylate synthase (TS) mRNA concentrations, thymidine phosphorylase, and the dihydropyrimidine dehydrogenase. The three genes involved can be used to predict fluorouracil(Drug information on fluorouracil) (5-FU) response and to customize adjuvant chemotherapy by using other available drugs, such as irinotecan, for patients expected to be fluorouracil resistant.[10,11]

Preliminary data suggest that gene expression levels of topoisomerase 1, p21, bcl2 may predict response to irinotecan therapy. Also, increased toxicity in patients treated with irinotecan may be explained by polymorphism in the UGT 1A1 gene. Recently, a new method for isolating mRNA and measuring TS gene expression from paraffin(Drug information on paraffin)-embedded tissue section using TaqMan technology has been developed at the University of Southern California.[12] TS gene expression was also shown to influence the response to 5-FU-based chemotherapy and survival in patients with gastric cancer.[13]

Fluorouracil vs. Irinotecan

Trials have begun to be designed in which patients with intratumoral relative TS gene expression higher than 4´ 10^-3 are treated with irinotecan while patients with lower than 4´ 10^-3 are treated with 5-FU.[12] In fact, there are many common similarities between colorectal cancer and non-small-cell lung cancer. They have some seemingly equal genetic profiles characterized by multiple losses of heterozygosity (LOH) on chromosomes 5q, 17p (in the p53 region, among others). For example, the adenomatous polyposis coli (APC) gene is methylated at a very high frequency (> 95%) in esophageal adenocarcinoma. Those patients with high levels of methylated APC in their pretreatment serum had a significantly shorter survival time than patients with undetectable or low levels.

More importantly, analysis of several serum specimens obtained from several patients under treatment show that serum status of the marker correlated with the degree of tumor response. The Danenbergs have reported that lung tumors also harbor nearly a 95% frequency of methylated APC genes. The frequency of detection of methylated APC in the serum of lung cancer patients was 50%. Analysis of serum samples for stage I lung cancer patients show that methylated APC levels predicted treatment outcome.[14]

11p15.5 LOH as Potential Genetic Surrogate Marker

Several markers of lung cancer susceptibility have been identified.[15] We found a higher percentage of rare HRAS1 VNTR alleles in lung cancer patients than in unaffected controls.[16] In addition, lung cancer cells contain deletions of multiple chromosomal regions. Deletion mapping studies have defined more than 30 regions dispersed in 21 different chromosome arms as candidate tumor suppressor loci.[17]

Polymorphic DNA probes for chromosome 3p were used to compare tumor and constitutional genotypes of nine small-cell lung cancer patients, with loss of heterozygosity (LOH) a common feature in tumor DNA of all patients.[18] Since then, allele loss involving chromosome arm 3p is one of the most frequent and earlier known genetic events in lung cancer. Intriguingly, in a recent chromosome 3p allelotyping it was noted that most small-cell lung carcinomas (91%) and squamous cell carcinomas (95%) exhibited larger 3p segments of allele loss, whereas most (71%) of the adenocarcinomas had smaller chromosome, nearer 3p allele loss.[19]

Evaluating LOH

In our recent experience (unpublished) evaluating loss of heterozygosity, we used primers flanking dinucleotide microsatellite repeat polymorphisms spanning the chromosome 3p arm, which is located at chromosome 3p14 (D3S1285) and marker D3S1234 and at chromosome 9p21 (D9S1747) and p53 in specimens of resected NSCLC patients and in their serum DNA. We analyzed the data with customized software designed to facilitate color formatting and visualization of data. Microsatellite analysis for LOH in 54 resected NSCLC patients is depicted in Figure 1. 

For another excellent review about tumor suppressor genetic abnormalities in lung cancer, see Pitterle et al.[20] Beyond the potential prognostic value of LOH, especially in stage I patients, these numerous chromosomal deletions could be involved in drug resistance.

Recently, Bepler et al[21] reported that 11p15.5 LOH, including the gene for the enzyme ribonucleotide reductase, predicts poor survival. Survival of patients with stage I and LOH (23 patients) is equal to that of patients with stage II disease, whereas survival significantly improved (P = .003) in patients with stage I and no LOH (101 patients). The LOH 11A region is lost in 75% of lung cancers. This region is located on chromosome segment 11p15.5 between the b-hemoglobin (b-hhh) and insulin genes.[22] A number of imprinted genes with maternal expression have been identified at 11p15: BWR1A (or IMPT1/TSSC5), BWRIC (or IPL/TSSC3), and CDKNIC, which encodes the cyclin-dependent kinase inhibitor p57^KIP2 that is found mutated in some cases of Beckwith-Wiedemann syndrome (an excessive-growth genetic disorder that predisposes to Wilm’s tumor).

Gene Clustering

A striking feature of imprinted genes in mammals is their tendency to cluster in the genome. A major imprinted gene cluster occurs in human chromosome 11p15.5. The centromeric and telomeric boundaries of the LOH 11A region is defined for the markers D11S988 and D11S860, which are spaced at an interval of 580-655 kb; the LOH 11A region contains the genes SSA/RO52, RRM1, and GOK.[21] Bepler et al[22] constructed a precise physical map containing the b-hemoglobin gene cluster region through the gene for ribonucleotide reductase subunit M1(RRM1) to the marker D11S860. They mapped the location of 6 polymorphic markers, 22 potential genes, and 18 transposable elements.

Previous studies of the Bepler group[23,24] show that LOH 11A (allele loss) is associated with metastasis formation in patients with lung cancer.[23] This unusual finding is explained by the presence of the RRM1 gene in the LOH 11A region that encodes the large subunit (M1) of the enzyme ribonucleotide reductase,[24] which is involved in the metabolism of gemcitabine.[25] Hence, we can speculate that reduced ribonucleotide reductase activity influences the dNTP pools that in turn can influence the incorporation of gemcitabine into DNA. The RRM1 gene occupies only 55 kb of the LOH region and additional genes may be involved in lung cancer.