Histopathologic and genetic alterations as predictors of response to treatment and survival in lung cancer: A review of published data

3.1.1. Histopathologic characteristics 

Squamous-cell carcinoma (SCC) originates from bronchial epithelial cells in the larger bronchi (lobular and segmental) and is frequently located centrally, but in up to 25% of cases it can be detected in peripheral bronchioli. It usually forms a solid mass with intrabronchial nodular development, intraepithelial and local expansion with infiltration of adjacent structures, whereas distant metastases are less frequent compared to adenocarcinomas and occurs late in the natural course of the disease. Well-differentiated SCC may be accompanied by formation of keratin, keratinous intercellular bridges and central necrosis can be found in 5% of cases. Consequently, the majority of SCC expresses high-molecular weight keratins (34βE12), cytokeratins 5/6 and the P63 gene and protein [10]. For the time being, no specific histopathologic variables have been reported to be of prognostic value in SCC and thus TNM stage and performance status of the patient remain the most important prognostic clinicopathological parameters. Notably, grade of histological differentiation seems to possess independent prognostic value in multivariate analysis in almost all clinical trials regarding SCC [2], [4].

3.1.2. Molecular prognostic markers 

SCC is characterized by overexpression of genes encoding cytokeratins, a finding which is consistent with its histopathologic findings of keratinous differentiation [11]. The use of comparative genomic hybridization assays has shown that genes coding for cytokeratins 5, 6, 13, 14, 16, 17 and 19 are amplified in the vast majority of SCC cases [12]. Importantly, 84% of SCC cases overexpress the epidermal growth factor receptor (EGFR) gene [13]. Measurable levels of EGF are found in SCC more often than in other NSCLC subtypes [14]. HER-2/neu expression, which is frequent in adenocarcinomas, is detected only occasionally in SCC. Similarly, K-ras mutations are also infrequent in SCC, even in smokers [3].

Several molecular prognostic markers have been proposed for SCC, including reduced expression of the cyclin-dependent kinase inhibitors p16INK4 and p2WAF [8], overexpression of cyclins D and E [9] and inactivation of the tumor-suppressor genes RB1, FHIT and TP53 [5]. A meta-analysis of 43 studies showed that TP53 mutations or overexpression are inversely related to prognosis in cases of ADC but not in SCC [15]. On the contrary, loss of PRb expression is predictive for poor survival in SCC [7]. An illustration of molecular changes leading to neoplastic transformation in SCC is provided in Fig. 1.

View Large Image Download to PowerPoint

Fig. 1. Serial genetic events provoking neoplastic transformation in alveolar/broncho-alveolar and bronchial epithelial cells.

3.2.1. Histopathologic characteristics 

Adenocarcinoma of the lung is a malignant epithelial tumor characterized by adenomatous differentiation with a papillary, nodular, alveolar or broncho-alveolar histopathologic pattern with or without mucous production or a combination of the above. Although most of the cases occur in smokers, ADC is the most common subtype among never-smokers, especially women [3]. Macroscopically, adenocarcinomas appear most frequently as peripheral tumors, either as intrabronchial tumors or as diffuse “pneumonia-like” infiltrations often involving both lungs and even the parietal pleura. Mixed ADC is the most common histologic subtype representing almost 80% of resected ADC [16].

Immunohistochemical characteristics of ADC vary according to the histologic subtype and grade of differentiation. Expression of epithelial markers, including AEI/AE3, CAM 5.2, EMA and CEA is typical. Cytokeratin 7 is expressed more often than cytokeratin 20 [17], while surfactant apoprotein A and product of type 2 alveolar cells, are overexpressed in up to 60% of ADC cases. Thyroid transcription factor 1 (TTF-1) is very frequently expressed, especially in well-differentiated tumors. However, broncho-alveolar carcinomas (BAC), especially when accompanied by mucous production, are often TTF-negative and positive for cytokeratins 7 and 20 [18].

3.2.2. Histopathologic prognostic factors 

Histological parameters that have been correlated with poor prognosis in ADC include low differentiation (high-grade tumors) and vascular infiltration [19]. Increased mitotic activity [19], poor tumor-associated lymphocyte infiltration [20] and extensive tumor necrosis [21] have also been reported as adverse prognostic markers. The presence of stromal reaction in small (less than 2cm) adenocarcinomas is of significant prognostic importance. Patients with small broncho-alveolar adenocarcinomas without central desmoplastic reaction have been reported to have 10-year survival of almost 100% [22].

3.2.3. Molecular prognostic markers 

Genetic changes during neoplastic transformation in ADC include point mutations in oncogenes, like KRAS, and in tumor-suppressor genes, like TP53 [23] and p16INK4 [24]. Mutations in KRAS have been detected in up to 30% of ADC, especially in smokers, and include mainly point mutations in exons 12 and 13 [25]. Mutations in TP53 have been reported to be an independent adverse prognostic factor in early stage (I and II) ADC [25]. Increased expression of the negative cell-cycle regulator p27 has been correlated with well-differentiated tumors and favorable prognosis [26]. COX-2 overexpression has also been reported in lung ADC but its prognostic significance is undetermined [27].

To date, no widely accepted molecular prognostic markers exist for lung ADC that could be implemented in every-day clinical practice. In some recent reports, KRAS oncogene activation with point mutations and HER-2/neu overexpression have been reported to correlate with poor survival. Some researchers suggest that the combined use of the three molecular markers (p53, KRas and c-erbB2) can contribute to prognostic evaluation [28]. More recently, global gene expression analysis using microarrays identified distinct gene expression profiles corresponding to different histological subtypes of ADC with prognostic significance [29], [30]. An illustration of molecular changes leading to neoplastic transformation in ADC is provided in Fig. 1.

3.3.1. Histopathologic characteristics 

Large-cell lung cancer (LCLC) is a poorly differentiated non-small-cell lung carcinoma that is not characterized by adenomatous or squamous differentiation. In most series, LCLC represents 9% of all lung cancers among which 3% exhibit evidence of neuroendocrine differentiation. By definition, diagnosis of LCLC requires the absence of adenomatous or squamous morphological or immunohistochemical evidence. Diagnosis of LCLC with neuroendocrine differentiation requires the use of immunohistochemical markers including chromogranin A, synaptophysin and NCAM (CD56). Approximately half of these tumors also express TTF1 [31].

3.3.2. Histopathologic prognostic factors 

Immunohistochemical evidence of neuroendocrine differentiation in LCLC has not been proven to be an independent prognostic factor per se. Several studies concluded that there is no difference in the prognosis of large-cell carcinomas with or without neuroendocrine elements [32], [33]. LCLC are often characterized by histological and genetic alterations similar to those described in other types of NSCLC and they represent the least differentiated histologic subtype compared to adenocarcinomas and squamous-cell carcinomas. At the same time, they share common pathogenetic characteristics with SCLC and other neuroendocrine tumors of the lung, including carcinoids. The lymphoepithelioid subtype of LCLC is characterized by the presence of Ebstein-Barr virus (EBV) DNA sequences which indicate the virus-dependent process of carcinogenesis in epithelial bronchial and alveolar cells in this tumor subtype and is associated with poor prognosis [33].

3.3.3. Molecular prognostic markers 

Large-cell neuroendocrine carcinoma harbors specific genetic alterations similar to those of SCLC, including LOH in 3p22-24 and point mutations in FHIT, TP53 and RB1 [34]. TP53 LOH and point mutations have been reported to correlate with poor prognosis [35]. Other patterns of molecular changes resemble the process of malignant transformation in undifferentiated NSCLC, including point mutations in KRAS and TP53 genes, overexpression of cyclins D1 and E and reduced expression of p16INK4 and RB1 with the same frequency as in other subtypes of NSCLC. Among all these molecular changes, LOH and point mutations in TP53 are the only ones that have been shown consistently to be independent prognostic factors for poor survival [35].

An overview of the main histopathologic and genetic alterations reported in NSCLC in relation to histological subtypes is provided in Table 1, Table 2, respectively. Table 3 offers a summary of the main molecular changes involved in NSCLC carcinogenesis for which prognostic significance has been evaluated and reported. A value of HR>1 suggests that high expression leads to decreased survival whereas HR<1 means that high expression leads to decreased survival.

Table 1. Histopathogenetic alterations in lung cancer in relation to tumor histology.

	Histopathologic event	SCC	ADC	LCLC	SCLC	Reference
	Keratin formation	60–80%	<5%	<5%	<1%	[10], [11]
	Surfactant A protein expression	No	60%	No	No	[17]
	P63 protein expression	Yes	No	No	No	[10]
	TTF-1 protein expression	Never	Always (except BAC)	50%	Never	[18], [31]
	HER-2 protein expression	15–25%	15–35%	NR	NR	[69]
	EGFR protein expression	80–85%	40–60%	NR	5-15%	[61], [62], [63], [64], [65]
	Neuroendocrine differentiation	<5%	<5%	15–35%	Always	[31], [32], [33], [34], [35]
	ERCC1 protein expression (IHC)	55%	29%	NR	10–25%	[36], [37], [38], [39], [40], [41]
	Cyclin D, E-2F expression	Increased in 10%	Increased	Increased	Significantly increased	[8], [9]
	pRb protein expression	Reduced	Reduced in 15%	Reduced	Reduced in most cases (>80%)	[8], [9]

Abbreviations: SCC, squamous-cell carcinoma; ADC, adenocarcinoma; LCLC, large-cell lung cancer; SCLC, small-cell lung cancer; TTF1, thyroid transcription factor 1; EGFR, epidermal growth factor receptor; ERCC1, excision-repair cross-complementation group 1; IHC, immunohistochemistry; NR, not reported.

Table 2. Main genetic changes in lung cancer in relation to tumor histology.

	Molecular event	SCC	ADC	LCLC	SCLC	Reference
	LOH 3p	Up to 80%	Up to 80%	50–60%	Up to 80%	[3], [5], [34], [35]
	P53 gene mutation	Mutated in up to 40%	Mutated in 30%	Mutated in 40%	Rarely mutated	[5], [6], [7], [34], [35]
	P63 gene expression	Yes	No	No	No	[10]
	Cytokeratine gene expression	CK5, 6, 13, 14, 16, 17, 19	CK7, CK20	No	No	[12], [17]
	EGFR point mutations	Infrequent	Frequent (40%) in smokers	Infrequent	Infrequent	[48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [72]
	EGFR gene expression	30–35%	45–55%	NR	<5%	[13], [57], [60], [61], [62]
	K-ras mutations	Infrequent	15–30% in smokers	Frequent in smokers	Infrequent	[3], [25], [48], [72]
	P16INK4 expression	Reduced in more than 70%	Reduced	Reduced	Reduced in 50%	[8], [9]
	P21WAF expression	Reduced	Reduced	NR	NR	[3]
	Cyclin D, E-2F expression	Increased in 10%	Increased in 25–40%	Increased	Significantly increased (>70%)	[8], [9]
	pRb gene expression	Reduced in 15–35%	Reduced in 15–35%	Reduced	Reduced in 80–100%	[8], [9]
	COX-2 gene expression	Increased	Increased	NR	Reduced	[27]
	Chromogranin gene expression	No	No	90%	90–100%	[34], [35]
	c-Myc gene amplification	Infrequent	Infrequent	Infrequent	Frequent	[3], [8]
	Bcl-2 gene expression	Normal	Normal	Normal	Increased	[8], [9]

Abbreviations: SCC, squamous-cell carcinoma; ADC, adenocarcinoma; LCLC, large-cell lung cancer; SCLC, small-cell lung cancer; TTF1, thyroid transcription factor 1; EGFR, epidermal growth factor receptor; ERCC1, excision-repair cross-complementation group 1; IHC, immunohistochemistry; NR, not reported.

Table 3. Main molecular prognostic markers in NSCLC.

	Marker	Number of studies	Total patient number	Technique	Hazard ratioa	References
	Cyclin A	2	302	IHC	1.73	Volm et al. (1997)
						Dobashi et al. (2003)
	Cyclin B1	2	156	IHC	10.7	Soria et al. (2000)
						Yoshida et al. (2004)
	Cyclin D1	4	475	IHC	0.21–3.93	Nishio et al. (1997)
						Keum et al. (1999)
						Gugger et al. (2001)
						Jin et al. (2001)
	Cyclin E	5	676	IHC+RT-PCR	1.3–1.92	Fukuse et al. (2000)
						Mishina et al. (2000)
						Hayashi et al. (2001)
						Muller-Tidow et al. (2001)
						Dobashi et al. (2003)
	pRb	5	1024	IHC+Northern	0.26–0.74	Reissmann et al. (1993)
						D’Amico et al. (1999)
						Jin et al. (2001)
						Akin et al. (2002)
						Haga et al. (2003)
	p16	8	844	IHC+Western	0.03–0.53	Kratzke et al. (1996)
						Taga et al. (1997)
						Groeger et al. (1999)
						Kawabuchi et al. (1999)
						Huang et al. (2000)
						Jin et al. (2001)
						Gonzalez et al. (2002)
						Esposito et al. (2004)
	p21	3	438	IHC	0.55–0.59	Komiya et al. (1997)
						Shoji et al. (2002)
						Esposito et al. (2004)
	p27	4	474	IHC	0.23–0.59	Esposito et al. (1997)
						Hommura et al. (2000)
						Hayashi et al. (2001)
						Tsukamoto et al. (2001)
	Bcl-2	7	1277	IHC	0.19–0.76	Ohsaki et al. (1996)
						Higashiyama et al. (1997)
						Fontanini et al. (1998)
						Silvestrini et al. (1998)
						Laudanski et al. (1999)
						Cox et al. (2001)
						Han et al. (2002)
	P53 mutations	10	1320	DNA sequence	1.95–4.73	Horio et al. (1993)
						Mitsudomi et al. (1993)
						Vega et al. (1997)
						Fukuyama et al. (1997)
						Huang et al. (1998)
						Hashimoto et al. (1999)
						Tomizawa et al. (1999)
						Skaug et al. (2000)
						Laudanski et al. (2001)
						Ahrendt et al. (2003)
	VEGF expression	14	1474	IHC	1.5–7.09	Volm et al. (1997)
						Imoto et al. (1998)
						Fontanini et al. (1998)
						Giatromanolaki et al. (1998)
						Volm et al. (1998)
						Decaussin et al. (1999)
						Yuan et al. (2000)
						Koukourakis et al. (2000)
						Ohta et al. (2000)
						O’Byrne et al. (2000)
						Liao et al. (2001)
						Han et al. (2001)
						Kojima et al. (2002)
						Mineo et al. (2004)
	PDGF expression	3	422	IHC	0.68–4.14	Koukourakis et al. (1997)
						O’Byrne et al. (2000)
						Kojima et al. (2002)
	FGF expression	2	289	IHC	2.26–2.29	Takanami et al. (1996)
						Kojima et al. (2002)

a HR>1 means that high expression leads to decreased survival. HR<1 means that high expression leads to increased survival. Abbreviations: IHC, immunohistochemistry; RT-PCR, real-time polymerase chain reaction.

5.1.1. EGFR gene mutations 

Correlation between somatic mutations of the EGFR gene and dramatic clinical response to the EGFR-TKI gefitinib was reported on-line for the first time in April 2004 by two groups [53], [54]. Thirteen out of 14 patients with EGFR gene mutations experienced objective responses to treatment with gefitinib, whereas none of the 11 tumors in non-responders harbored any EGFR mutations. Interestingly enough, all mutations were located at the tyrosine kinase domain of EGFR [53], [54].

Subsequent studies suggested that EGFR mutations are detected more often in women, never-smokers, patients of Asian origin and in ADC histology [52], [55]. Due to these characteristics, and especially the non-smoking status, patients with tumors harboring EGFR mutations are considered to have better prognosis independently of treatment selection. In a randomized clinical trial comparing chemotherapy alone or in combination with erlotinib in patients with advanced ADC, patients with tumors carrying EGFR mutations had a better clinical outcome independently of the treatment arm [52]. In a similar way, patients with EGFR mutations who participated in the BR.21 study had better survival even when allocated in the placebo arm (HR=0.70), confirming thus the independent prognostic value of EGFR mutations for survival in lung ADC patients [56].

EGFR mutation types in lung ADC are the same worldwide and consist of multiple deletions in exon 19 of the gene in 45% of cases, missense point mutations in exon 21 in 40% of cases (mainly the L858R transition) and missense or insertion mutations in exons 18–21 in 15% of the cases [57], [58], [59], [60], [61]. No robust scientific evidence exists to date associating EGFR-TKI activity to the type of mutation. Acquired resistance to TKIs has been recently associated to an additional mutation (T790M) affecting the ATP pocket of the intracellular EGFR-tyrosine kinase domain impairing thus TKI binding [62]. Better clarification of these molecular mechanisms and their predictive value may lead to the development of “2nd” generation EGFR-TKIs that will target the additional T790M mutation [63]. Interestingly, patients with NSCLC of other histological subtypes and even patients with lung tumors harboring “wild type” EGFR may derive substantial clinical benefit from EGFR-TKIs. It is therefore important to further investigate how to best incorporate EGFR genetic screening in clinical decision making.

5.1.2. EGFR gene copy number 

A higher number of EGFR gene copies evaluated by fluorescent in situ hybridization (FISH) has been reported to correlate with worse clinical outcome compared to patients with tumors without EGFR gene amplification [61], [65]. In these studies, FISH positivity was defined as true gene amplification or high polysomy with more than 4 EGFR gene copies in more than 40% of cancer cells. The randomized BR.21 trial [56] showed that EGFR FISH(+) patients (approximately 40% of the whole study population) who were allocated to the erlotinib treatment arm had significantly better survival compared to FISH(+) patients who received placebo (HR=0.44, p=0.01). In FISH(−) patients there was no difference in survival in relation to the treatment arm (HR=0.93). In the Iressa® Survival Evaluation in Lung Cancer (ISEL) study, FISH(+) patients (approximately 30% of the whole study population) who received gefitinib had prolonged survival compared to FISH(+) patients who were allocated to the placebo arm (HR=0.61, p=0.06). Again, in FISH(−) patients there was no benefit in overall survival for patients allocated to the gefitinib arm (HR=1.16, p=0.42). Importantly, in the ISEL study, FISH(+) patients treated with gefitinib had a better prognosis independently of sex, histology or smoking status [66]. The above mentioned data support an important predictive role of EGFR evaluation for response to treatment with EGFR-TKIs, potentially similar to that of HER-2 evaluation in breast cancer.

5.1.3. EGFR gene polymorphisms 

The EGFR gene contains a particularly polymorphic nucleotide sequence in intron-1, consisting of a varying number (15–22) of repeated dinucleotide sequence CA (cytosine–adenosine). The allele 16 (containing 16 repeats) is the most frequent (42%), followed by the alleles 20 (26%) and 18 (20%) [67]. In vitro and in vivo studies have shown that, due to DNA binding site modifications, changes in this domain which acts as the promoter region affect transcriptional capacity in such a way that the bigger the number of repeats, the lower the levels of EGFR mRNA and of EGFR protein expression [68], [69]. For example, the 22-repeat allele causes 80% reduction of gene expression compared to the 16-repeat allele [68]. Based on these data, response to treatment of NSCLC with EGFR inhibitors may vary among patients due to racial genotypic differences. More specifically, the high response rate observed in Japanese patients with EGFR-targeted agents may result from low EGFR expression due to the long 20-repeat allele of the intron-1 region of EGFR that is the most common in Asia, whereas the shorter 16-repeat allele is most common in the white and black race [70]. Nevertheless, it should be noted that EGFR expression has more consistently been correlated with toxicity rather than the efficacy of EGFR inhibitors, such as in the case of skin rash which is the most frequent toxicity associated with these agents [71].

In NSCLC, DNA sequencing analysis for EGFR mutations or polymorphisms in patients with stage II-IIIA disease who received adjuvant chemotherapy and radiotherapy in the context of the ECOG 3590 study, has shown that patients with tumors carrying more than 35 CA repeats in the intron-1 of the EGFR had significantly longer median survival (41 months) compared to those with 35 or less (29.2 months) [72]. Nevertheless, it should be noted that this study was not planned to examine this polymorphism as a prognostic factor by itself or as a predictive factor for response to treatment with EGFR-TK inhibitors. In contrast, Han et al. and Liu et al. have found that the small number of CA repeats is associated with better response to gefitinib treatment in patients with NSCLC [73], [74]. Moreover, in patients with NSCLC, squamous carcinoma of the head and neck (SCCHN) and ovarian cancer treated with erlotinib, it was reported that dermatologic toxicity is associated with intron-1 CA repeat variability [75].

5.1.4. EGFR protein expression 

Conflicting results have been reported to date concerning the prognostic significance of EGFR protein expression. In the IDEAL 1 and 2 studies, retrospective immunohistochemical (IHC) evaluation of EGFR expression failed to show any correlation between EGFR protein levels and response to treatment with gefitinib [76]. Using a different IHC evaluation protocol (scale 0–400) and a different monoclonal antibody, Hirsh et al. and Capuzzo et al. studied the relationship between EGFR IHC positivity and clinical outcome after treatment with gefitinib. Both studies demonstrated that ICH(+) patients had significantly higher objective responses, time to progression and survival compared to patients with low ICH score as defined by the protocol criteria [77], [78]. In the BR.21 study, IHC(+) patients (57% of the whole study population) treated with erlotinib had better survival compared to IHC(+) patients in the placebo arm (HR=0.68, p=0.08) [56]. In the ISEL trial, IHC(+) patients allocated to the gefitinib treatment arm had prolonged survival compared to IHC(+) patients treated with placebo, albeit not statistically significant (HR=0.77, p=0.13) [66].

An important question is whether the combined use of two tests could yield a better predictive and/or prognostic result compared to one test. Hirsch et al. classified their patients in three groups according to positivity for increased EGFR gene copy number (evaluated by FISH) and EGFR protein expression (evaluated by IHC): (a) FISH(+)/ICH(+), (b) FISH(+)/IHC(−) or FISH(−)/ICH(+) and (c) FISH(−)/ICH(−). They found that objective responses after treatment with EGFR-TKIs were 41%, 10% and 2%, respectively, and median survival declined from 21 months for the “double positive” patients to 11 months for the “single positive” group and to only 6 months for “double negative” patients [77]. Notably, median survival of the “double positive group” was significantly superior to that reported with the most active chemotherapy combinations in NSCLC, including the paclitaxel, carboplatin and bevacizumab triplet.

5.2.1. Vascular endothelial growth factor (VEGF) in NSCLC 

Vascular endothelial growth factor (VEGF) is a potent growth factor for endothelial cells. Its expression is stimulated by tissue hypoxia, as well as several different growth factors and cytokines. It binds to its receptors, VEGFR-1 and VEGFR-2 and to a lesser extent to VEGFR-3, causing proliferation and migration of endothelial cells, promoting thus angiogenesis [79]. VEGF also increases vascular permeability and may be involved in the coagulation, fibrinolysis and apoptosis pathways. In contrast to many other markers studied in NSCLC, increased VEGF expression has consistently been shown to adversely affect NSCLC outcome (Table 3). Bevacizumab (Avastin®) is a humanized monoclonal antibody that inhibits binding of the vascular endothelial growth factor A (VEGF-A) to its receptors VEGFR-1 and VEGFR-2 [80]. Bevacizumab has been approved for the first line treatment of metastatic or inoperable NSCLC [81]. Although VEGF overexpression has been directly associated with the process of angiogenesis in NSCLC, the correlation between VEGF levels and response to treatment with bevacizumab has not yet been adequately defined [82].

More than 30 SNPs have been recognized to date in the VEGF gene [83]. The VEGF C396T SNP in the 5′-UTR region of the gene has been correlated with low plasma levels of VEGF [84]. In NSCLC, VEGF polymorphisms have been correlated with overall survival. More specifically, in 462 patients with respectable NSCLC, the SNPs VEGF 936C>T, VEGF 460T>C and 405G>C were evaluated. It was reported that patients with VEGF 405C allele had significantly longer survival (HR=0.70, 95% CI: 0.54–0.91, p=0.008), while the VEGF 936T allele tended to correlate with overall survival as well (HR=0.73, p=0.07) [85]. Given the fact that bevacizumab is currently under intensive testing as a potential therapeutic agent in types of lung cancer, including SCLC, information related to possible impact of numerous polymorphisms on clinical efficacy or toxicity is likely to be of particular interest in the near future.

5.2.2. Platelet-derived growth factor (PDGF) and fibroblast-growth factor (FGF) in NSCLC 

Platelet-derived growth factor (PDGF) is secreted from platelets and increases DNA synthesis, endothelial cell migration and tumor growth [79]. The impact of PDGF levels on clinical outcome remains unclear with two studies showing no effect, whereas one study limited to stage I disease showed poorer survival in PDGF-expressing tumors (Table 3). Fibroblast-growth factor (FGF), on the other hand, is released by proteolytic enzymes from the extracellular matrix, after which it increases the expression of other proteolytic molecules [79]. Limited studies in lung cancer suggest that FGF expression is an independent prognostic marker in patients with ADC, whereas a single study in patients with SCC did not report a statistically significant effect on survival (Table 3). Finally, hepatocyte growth factor (HGF), a cytokine produced by mesenchymal cells regulates function of epithelial and endothelial cells through its receptor, c-met protein. Limited and conflicting data on HGF and c-met suggest that these markers may be of prognostic value in NSCLC [86], [87], [88].

5.4.1. Gene expression profiling and prognosis 

The aim of these studies is to identify gene expression profiles from global analysis that predict disease-free survival and overall survival. Accordingly, these profiles should be able to outperform current clinicopathological prognosticators, for example by distinguishing different prognoses in patients with the same histological type and clinical stage of tumor [80].

Several studies have attempted to identify gene expression profiles involving between 22 and 50 genes that can separate otherwise identical tumors in terms of survival after potentially curative resection [98], [99], [100], [101]. Although these profiles could form the basis for individualized therapeutic selection involving the rational use of adjuvant or neoadjuvant systemic therapies, all studies to date refer on relatively small samples and involve retrospective analysis, with some overlap seen between profiles produced between different groups [80]. Other studies have attempted to identify a gene expression profile that “detects” tumors with “high metastatic potential” and thus identify patients with a high probability of micrometastatic disease. This approach has been used to evaluate the differences between the gene expression profiles in tumors of indistinguishable histological type and clinical stage that do and do not metastasize [99].

Based on this data, Chen et al. identified sixteen genes that correlated with survival among patients with resectable NSCLC by analyzing microarray data and risk scores and subsequently selected five genes for RT-PCR and decision-tree analysis to develop a five-gene signature that was an independent predictor of relapse-free and overall survival [102]. This signature can be useful in stratifying patients according to risk in trials of adjuvant treatment of the disease.

5.4.2. Gene expression profiling and response to treatment 

There are few studies examining gene expression profiling in relation to cytotoxic therapy in NSCLC [49], [103], [104]. Nevertheless, large databases of gene expression profiles that determine the sensitivity or resistance to particular cytotoxic or the newer “targeted” agents have been developed, based on studies of a large number of agents in a large number of cancer cell lines [105], [106], [107]. Importantly, these profiles appear to be specific for the mechanism of action of the drugs with similar pharmacology, in a way that drugs with common properties and action cluster together within the same profile. Using these data, a number of important signaling pathways that are involved in carcinogenesis of the lung have now been elucidated through informed platforms for assessing biomarkers associated with sensitivity or resistance to various agents targeting these pathways [108]. Furthermore, the gene expression profiles for each antineoplastic drug can be applied to predict the chemosensitivity or resistance of a particular cell line to that drug with a high degree of accuracy [105].

Importantly, the use of gene expression profiling in time-course studies with multiple cell lines has allowed evaluation of the changes in gene expression that occur after cytotoxic drug exposure and provided useful insights into molecular determinants of drug resistance after expose to chemotherapy. The genes identified by these methods may prove to be useful targets for novel drug development or could act as senescence-specific markers for future studies in cases of chemotherapy-induced senescence in malignant cell lines [109].