Notch signaling plays a key role in embryonic vascular development and angiogenesis. The authors aimed to study the prognostic role of the angiogenesis-related Notch ligands and receptors and investigate the prognostic impact of the coexpression of vascular endothelial growth factor-A (VEGF-A) and Notch signaling.
Tumor tissue samples from 335 resected patients with stage I to IIIA nonsmall cell lung cancer (NSCLC) were obtained, and tissue microarrays were constructed from duplicate cores of tumor cells and tumor-related stroma from each specimen. Immunohistochemistry was used to evaluate the expression of the molecular markers Notch-1, Notch-4, Delta-like ligand 4 (DLL4), and Jagged-1.
There were 191 squamous cell carcinomas (SCCs), 113 adenocarcinomas (ACs), and 31 large cell carcinomas. In AC, low tumor cell Delta-like ligand 4 expression was an independent negative prognostic factor (hazard ratio [HR], 2.9; 95% confidence interval [CI], 1.4-6.3 [P = .006]), whereas high tumor cell Notch-1 expression was an independent negative prognostic factor (HR, 2.2; 95% CI, 1.2-4.1 [P<.001]). In SCC, low stromal Delta-like ligand 4 expression was an independent indicator of poor prognosis (HR, 3.3; 95% CI, 1.8-6.1 [P<.001]). The coexpression of Notch-1 and VEGF-A had a significant prognostic impact (P<.001). For Notch-1 and VEGF-A, low/low (n = 142) versus high/high (n = 35) expression resulted in 5-year survival rates of 69% and 32%, respectively.
Lung cancer is the number 1 cause of cancer-related death worldwide.1 Despite several new treatment achievements, the overall survival for lung cancer patients is disappointing.1, 2 As a consequence, there is a need for new treatment strategies.
Recently, there has been an increased focus on Notch signaling as a new promising target in cancer treatment, alone or in combination with, for example, angiogenic inhibitors. In addition to angiogenesis, inappropriate activation of Notch signaling results in increased proliferation, restricted differentiation, and prevention of apoptosis in cancer cells.3 There are 4 Notch receptors (Notch-1, 2, 3, and 4) and 5 known ligands, Jagged-1, Jagged-2, and Delta-like ligand-1, 3, and 4. The extent to which these ligands and receptors are involved in tumor angiogenesis is still under investigation, but at least Delta-like ligand 4, Jagged-1, Notch-1, and Notch-4 seem to have central roles.4-6 Activation of Notch signaling is initiated by ligand binding to Notch receptors mainly between adjacent cells, and there seems to be an important Notch cross-signaling between tumor, stromal, and endothelial cells.4, 6
As the prognostic impact of Notch signaling in different malignancies has been diverging, the Notch signaling role in cancer seems to be highly context and tissue specific and may be both oncogenic and tumor suppressive.7, 8 No previous lung cancer trials have, to our knowledge, investigated the potentially diverging prognostic impact of these markers in tumor versus stromal cells. However, evidence exists for a bidirectional communication involving Notch signaling between these cellular compartments in various malignancies.3, 9, 10 Hence, exploring the specific prognostic impact in nonsmall cell lung cancer (NSCLC) tumor versus stromal cells is of substantial interest.
Lung cancer is divided into small cell lung cancer (SCLC) (15%-20%), and NSCLC (80%-85%). The 2 major histological subtypes of NSCLC are squamous cell carcinoma (SCC) and adenocarcinoma (AC). During recent years, the treatment response and side effects by some new treatment options have been closely related to different NSCLC histologies. For instance, the vascular endothelial growth factor (VEGF) monoclonal antibody bevacizumab is only given to patients with AC because of the risk of fatal bleeding in SCC.11 Furthermore, the new antifolate agent pemetrexed appears to have better response in patients with non-SCC.12 As ACs and SSCs have been increasingly recognized as different diseases, in terms of both biology and treatment efficacy/side effects, we anticipate different impacts of Notch ligands and receptors in the various histological subgroups.
There is an increased use of antiangiogenic drugs, but the clinical benefits have been relatively modest. A more thorough understanding of the molecular and cellular mechanisms governing tumor angiogenesis is therefore pivotal.4 We have previously reported the prognostic impact of VEGF ligands and receptors in the same NSCLC cohort.13 There seems to be a close connection between Notch and VEGF activation, and a new treatment strategy blocking both these pathways has been tried in preclinical studies with success.4, 5 As a result, it is important to further assess the prognostic impact of the coexpression of Notch and VEGF-A.
In our unselected NSCLC cohort we 1) study the prognostic impact of Jagged-1, Delta-like ligand 4, Notch-1, and Notch-4 to assess whether they mediate different prognostic impact in tumor cells versus tumor-related stroma; 2) explore whether there are different impacts of Notch biomarkers in AC versus SCC; and 3) investigate whether the coexpression of Notch-1 and VEGF-A leads to a worse prognosis.
MATERIALS AND METHODS
Patients and Clinical Samples
Primary tumor tissues from anonymized patients diagnosed with NSCLC pathologic stage I to IIIA at the University Hospital of Northern Norway and Nordland Central Hospital from 1990 through 2004 were used in this retrospective study. In total, 371 patients were registered from the hospital database. Of these, 36 patients were excluded from the study because of: 1) radiotherapy or chemotherapy before surgery (n = 10), 2) other malignancy within 5 years before NSCLC diagnosis (n = 13), or 3) inadequate paraffin-embedded fixed tissue blocks (n = 13). Adjuvant chemotherapy had not been introduced in Norway during this period (1990-2004). Thus, 335 patients with complete medical records and adequate paraffin-embedded tissue blocks were eligible.
This report includes follow-up data as of November 30, 2008. The median follow-up of survivors was 86 months (range, 48-216 months). The tumors were staged according to the International Union Against Cancer TNM classification (sixth edition) and histologically subtyped and graded according to the World Health Organization guidelines.14 The National Data Inspection Board and the Regional Committee for Research Ethics approved the study.
All lung cancer cases were histologically reviewed by 2 pathologists (S.A-S. and K.A-S.), and the most representative areas of viable tumor cells (neoplastic epithelial cells) and central tumor stroma were carefully selected. The tissue microarrays (TMAs) were assembled using a tissue-arraying instrument (Beecher Instruments, Silver Springs, Md). The detailed methodology has been previously reported.13 Briefly, we used a 0.6 mm diameter stylet, and the study specimens were routinely sampled with 2 replicate core samples (different areas) of neoplastic tissue and 2 of tumor stroma. Both normal lung tissue localized distant from the primary tumor, and 1 slide with normal lung tissue samples from 20 patients without a cancer diagnosis, were used as controls. Multiple 4 μm sections were cut with a Micron microtome (HM355S) and stained by specific antibodies for immunohistochemistry analysis.
The applied antibodies were subjected to in-house validation by the manufacturer for immunohistochemistry (IHC) analysis on paraffin-embedded material. The antibodies used in the study were as follows: Delta-like ligand 4 (rabbit polyclonal; ab7280; Abcam, Cambridge, MA; 1:15), Jagged-1 (rabbit polyclonal; ab7771; Abcam; 1:100), Notch-1 (mouse monoclonal; ab3294; Abcam; 1:40), and Notch-4 (rabbit polyclonal; ab23427; Abcam; 1:100).
Delta-like ligand 4, Notch-1, and Notch-4 were stained using Ventana Benchmark XT (Ventana Medical Systems, Tucson, Ariz) with the Ultraview DAB procedure. Antigen retrieval was done automatically with CC1 mild (30 minutes).
Jagged-1 antigen retrieval was done manually by placing the specimens in 0.01 M citrate buffer at pH 6.0 and exposing them to microwave heating for 20 minutes at 450 W. The primary antibody was incubated overnight at 4°C. The marker was observed by adding a secondary antibody conjugated with biotin, followed by an avidin/biotin/peroxidase complex (Vectastein ABC Elite kit from Vector Laboratories, Burlingame, Calif).
For each antibody, including negative staining controls, all TMA stainings were performed in 1 single experiment.
Scoring of IHC
By light microscopy, representative and viable tissue sections were scored semiquantitatively for cytoplasmic staining. The dominant staining intensity in both tumor cells and stromal cells was scored as: 0 indicates negative; 1 indicates weak; 2 indicates intermediate; and 3 indicates strong staining intensity (Fig. 1). The cell density of the stroma was scored as: 1 indicates low density; 2 indicates intermediate density; and 3 indicates high density (Fig. 1). All samples were anonymized and independently scored by 2 pathologists (S.A-S. and K.A.-S.). In case of disagreement, the slides were re-examined, and a consensus was reached by the observers. In most tumor cores as well as in some stromal cores, there was a mixture of stromal cells and tumor cells. However, by morphological criteria we have only scored staining intensity of tumor cells in tumor cores and intensity and density of stromal cells in stromal cores.
The interobserver scoring agreement has previously been found valid in the same TMA blocks for 1 ligand and 1 receptor with similar cytoplasmic staining.13 Mean score for duplicate cores from each individual was calculated separately in tumor cells and stroma. High expression in tumor cells was defined as score ≥1.5 (Delta-like ligand 4), ≥2 (Jagged-1 and Notch-1), and >2 (Notch-4). For VEGF-A, we used the same cutoff value (>2) as previously published.13 Stromal expression was calculated by adding density score (1-3) and intensity score (0-3) before categorizing into low and high expression. High expression in stroma was defined as score ≥4 (Jagged-1, Delta-like ligand 4, and Notch-4). In contrast to the high variation of tumor cell staining of Notch-1, stromal cell staining showed almost no variation in intensity score between the cores, as all were weak positive. Consequently, stromal Notch-1 expression was not scored.
All statistical analyses were performed using the SPSS statistical package (version 15; SPSS, Chicago, Ill). The chi-square test and Fisher exact test were used to examine the association between molecular marker expression and various clinicopathological parameters. Plots of the disease-specific survival (DSS) according to marker expression were drawn using the Kaplan-Meier method, and statistical significance between survival curves was assessed by the log-rank test. DSS was determined from the date of surgery to the time of lung cancer death. Multivariate analysis was performed using the Cox proportional hazards model. Only variables of significant value from the univariate analysis were entered into the Cox regression analysis. Probability for stepwise entry and removal was set at .05 and .10, respectively. The significance level used was P<.05.
Demographic, clinical, and histopathological variables are shown in Table 1. The median age was 67 years (range, 28-85 years), and the majority of patients were male (75%). The NSCLC tumors comprised 191 SCCs, 113 ACs, and 31 large-cell carcinomas. Because of nodal metastasis or nonradical surgical margins, 59 (18%) patients received adjuvant radiotherapy.
Table 1. Prognostic Clinicopathologic Variables as Predictors for Disease-Specific Survival in 335 NSCLC Patients (Univariate Analyses; Log-Rank Test)
Expression of Notch Ligands and Receptors and Their Correlations
Notch ligands and receptors were expressed in the cytoplasm of tumor cells. On the basis of morphological criteria, most normal pneumocytes in control cores were negative or weakly stained by all antibodies. In tumor stroma and in control cores, most inflammatory cells (macrophages, lymphocytes, granulocytes, and plasma cells) were stained, and macrophages and plasma cells were strongly positive (except for Notch-1, where plasma cells were weakly stained in control cores). However, lymphocytes were mostly negative or weakly stained for Jagged-1. Regarding fibroblastlike cells, there was a variation in the staining intensity from weak to strong for all markers. Endothelial cells in both control cores and tumor-related stroma showed mostly negative to weak staining intensity for Notch-1, but were most often strongly stained for Jagged-1 and Delta-like ligand 4. Notch-4 expression in the endothelial cells showed various degrees of staining intensity in control cores, but was most often strongly positive in tumor stroma.
Notch ligands and receptors did not correlate with age, weight loss, differentiation, T-status, or vascular infiltration. In SCC, tumor cell Notch-1 expression (high expression: AC, 44%; large-cell carcinoma, 35%; SCC, 26% [P = .009]) and tumor cell Delta-like ligand 4 expression (high expression: large-cell carcinoma, 87%; AC, 86%; SCC, 73% [P = .004]) were significantly lower when compared with the other subgroups. Tumor cell Notch-1 was more frequently expressed in lymph node-positive patients (high expression: N0, 21%; N1, 32%; N2, 35% [P = .024]), whereas stromal Notch-4 was inversely associated with lymph node metastasis (high expression: N0, 19%; N1, 8%; N2, 8% [P = .01]).
Results from univariate analyses regarding clinical variables are shown Table 1. The influence on DSS by tumor cell and stromal expression of Notch ligands and receptors are shown in Table 2 and Figure 2. Results of the multivariate analysis are presented in Table 3. Among molecular markers, stromal Delta-like ligand 4 (hazard ratio [HR], 1.89; 95% confidence interval [CI], 1.19-3.00 [P = .007]) and tumor cell Notch-4 (HR, 1.55; 95% CI, 1.07-2.25 [P = .019]) expression appeared as independent prognosticators. For AC, tumor cell Delta-like ligand 4 (HR, 2.94; 95% CI, 1.37-6.32 [P = .006]) and Notch-1 (HR, 2.23; 95% CI, 1.21-4.11 [P = .01]) were independent prognostic factors, whereas for SCC, high stromal Delta-like ligand 4 expression (HR, 3.3; 95% CI, 1.79-6.09 [P<.001]) was a favorable independent predictor of prognosis.
Table 2. Tumor Cell and Stromal Notch Ligands and Receptors as Predictors of Disease-Specific Survival in NSCLC Patients (Univariate Analyses; Log-Rank Test)
We have previously presented the prognostic impact of VEGF-A in the same tumor material.13 Herein, we have used the same cutoff value, but because of a survival update in 2008 the follow-up time is prolonged. According to the tumor cell VEGF-A expression, the 5-year survival rates was 48% (high expression, n = 142) versus 66% (low expression, n = 192) (P = .001) after the survival update.
Figure 3 shows the Kaplan-Meier survival curves for the coexpression of Notch-1/VEGF-A in all patients (Fig. 3A), in SCC (Fig. 3B), and in AC (Fig. 3C). For all patients, 5-year DSS according to tumor cell coexpression levels of Notch-1 and VEGF-A were 69% (low/low, n = 142), 52% (low/high, n = 106), 55% (high/low, n = 46), and 32% (high/high, n = 35) (P<.001). In multivariate analysis, the 3-level coexpression of Notch-1 and VEGF-A data had an independent prognostic impact (P = .025). Between low/low and the combined high/low and low/high, the HR was 1.73 (95% CI, 1.13-2.67; P = .013), whereas between low/low and high/high the HR was 1.87 (95% CI, 1.05-3.34; P = .034).
We present a large-scale study in an unselected population of surgically resected NSCLC patients using high-throughput TMAs. We found diverging prognostic impacts of Notch biomarkers in AC versus SCC as well as different impacts of Notch biomarkers in tumor cells versus stromal cells. In AC, tumor cell Notch-1 was an independent poor prognostic factor. High Delta-like ligand 4 expression in AC tumor cells and SCC stromal cells was associated with a good prognosis. Finally, there seems to be a potentially interesting connection between Notch-1 and VEGF-A, as high tumor cell coexpression of Notch-1 and VEGF-A indicated a particularly poor prognosis.
Illustrating the diverging impact of Notch according to histological subgroups, we report, for the first time, Notch-1 as an independent negative prognostic factor in NSCLC patients with AC. Our findings corroborate previous results in other malignancies, such as breast15 and gastric cancer,16 but are in contrast to results from a colorectal study,17 where no association was found, and a bladder cancer study18 reporting high Notch-1 expression to indicate a good prognosis. In an IHC study of 153 NSCLC patients, Lee et al. reported Notch-1 to be negatively correlated to lymph node metastasis, but did not show any prognostic value.19 Although 51% of the tumors were ACs, there were no subanalyses with regard to histology. In a recent study of 420 NSCLC patients, Westhoff et al. did not observe a prognostic impact of Notch-1, but there was a tendency toward a negative prognostic impact in patients with high Notch-1 expression (P = .09).20 However, they used overall survival instead of DSS, and there were no subanalyses with regard to histology (58% AC). Nevertheless, in the subgroup of patients without p53 mutation (n = 176), they interestingly found Notch-1 to be an independent prognostic factor. In an in vitro lung cancer study, using the lung adenocarcinoma cell line A549, Zheng et al.21 concluded that Notch-1 functions as a tumor inhibitor. The same tumor suppressive effect by Notch-1 has previously also been found in SCLC.22 Supporting our results, Chen et al.7 observed that oxygen concentration determines the biological effect of Notch-1 signaling in lung AC. When AC cell lines were studied under standard tissue culture conditions, Notch-1 had a tumor suppressive effect, whereas under hypoxia (lung cancer tumors in vivo are hypoxic), Notch-1 signaling was required for AC cell survival.23 Notch signaling has previously been associated with tumor differentiation, but in the present study, there was no significant correlation between the Notch biomarkers and differentiation. However, future studies are needed to further explore the importance of Notch signaling, including downstream targets, such as Hey-1, Hey-2, and Hey-L.
It has been stated that Delta-like ligand 4 is expressed exclusively by endothelial cells.4 However, there seems to be an important Notch cross-signaling between tumor, stromal, and endothelial cells, and Martinez et al.24 recently demonstrated that Delta-like ligand 4 is widely distributed in tissues other than vessels. In agreement with our results, they found positive staining in normal lung tissue vessels and stroma, whereas pneumocytes were negative. They also showed positive staining in different types of tumors, including ⅔ of lung cancer epithelial cells. In an extensive work, Li et al.25 confirmed in 5 different xenografts (glioblastoma, prostate cancer, breast cancer, fibrosarcoma, and melanoma) that Delta-like ligand 4, also when expressed in tumor cells, functions as a negative regulator of tumor angiogenesis by reducing microvessel density. However, they found that fewer but larger vessels can be as efficient as a greater number of smaller vessels within the tumor. In human glioblastoma and prostate cancer, Delta-like ligand 4 acted as a positive driver for tumor growth, but not in breast cancer, fibrosarcoma, or malignant melanoma.
The positive prognostic impact of high tumor cell Delta-like ligand 4 expression in our NSCLC AC patients is in contrast to the negative prognostic impact of 1 of its main receptors, Notch-1, in the same subgroup. The same tendency was seen for stromal Delta-like ligand 4 expression in SCC. The discrepancy between the prognostic impact of receptor (Notch-1) and ligand (Delta-like ligand 4) may be explained by ligands other than Delta-like ligand 4 that are important for Notch-1 stimulation. Another possible explanation may be that the major prognostic impact of Delta-like ligand 4 is not related to angiogenesis. The mechanism behind the contradictory prognostic effects of Delta-like ligand 4 and Notch-1 is not clear, and further studies are needed to address this question.
In our study population, coexpression of Notch-1 and VEGF-A in tumor cells showed a prominent negative prognostic impact. This effect may be mediated through direct tumor cell proliferation, angiogenesis, or both. Bevacizumab (anti-VEGF monoclonal antibody) in combination with chemotherapy has been approved by the US Food and Drug Administration for patients with advanced nonsquamous NSCLC. Because of bleeding toxicity, patients with SCC have been excluded from bevacizumab-related NSCLC studies; consequently, the effect in this histological subgroup is uncertain. Anti-VEGF antibody in combination with Notch signaling inhibition appears as an interesting approach. Studies have shown that Notch-1 and Notch-4 can down-regulate VEGF receptor 2 (VEGFR2) expression.26, 27 This observation raises possible risks related to Notch inhibition, as it may lead to increased VEGFR2 signaling. Accordingly, a combination approach, for example, interruption of Notch signaling combined with VEGF inhibitors, may be required.6
In our study, patients with low VEGF-A and low Notch-1 expression had a clearly better prognosis than those with high expression of only 1 marker. This is consistent with murine studies showing that blockade of Notch signaling is also effective in growth inhibition of tumors resistant to VEGF-A inhibition.28 Among the patients with high Notch-1 and high VEGF-A expression (10%), the prognosis was very poor, but even more so in the AC subgroup. This result is consistent with evidence from xenograft tumors, in which combined blocking of both the Notch and VEGF axis appeared to have an additive effect.29 Although Notch and VEGF signaling are regarded as 2 of the most important angiogenic-related pathways, there is a wide range of other pathways interfering with tumor angiogenesis. Although promising, there are several possible pitfalls to overcome before this combination may be proven useful in NSCLC treatment. To our knowledge there is today, despite promising results from preclinical and some early phase clinical trials, no established use of Notch signaling targeted therapy in clinical cancer treatment.3, 5 Before clinical use, better knowledge about mechanisms, interaction, and effects has to be attained through further basic and translational research.5
These are the first published results in NSCLC on the prognostic impact of Notch biomarkers in coexpression with VEGF-A. In AC, tumor cell Notch-1 expression was independently associated with poor prognosis. In contrast, high tumor cell Delta-like ligand 4 expression in AC and high stromal Delta-like ligand 4 expression in SCC were predictive of a favorable prognosis. The coexpression of VEGF-A and Notch-1 led to a profoundly detrimental prognosis, supporting combined inhibition of VEGF/Notch signaling as an interesting approach in NSCLC.