Nitroglycerin Treatment May Enhance Chemosensitivity to Docetaxel and Carboplatin in Patients with Lung Adenocarcinoma

Purpose: Nitroglycerin may improve the response to chemotherapy in advanced non–small cell lung cancer. The effects and mechanisms of nitroglycerin on the enhancement of chemosensitivity to docetaxel and carboplatin regimen (DCb) in patients with lung adenocarcinoma have not been reported.

Experimental Design: Seventeen patients with operable lung adenocarcinoma and stable angina pectoris were selected to investigate the effects of nitroglycerin on immunoreactivity for hypoxia-inducible factor 1α (HIF-1α), vascular endothelial growth factor (VEGF), P-glycoprotein (P-gp), the production of which is regulated by HIF-1, and p53 proteins in their resected tumor by semiquantitative immunohistochemical analyses. Eight of 17 patients were treated with nitroglycerin patches before operation, but 9 of 17 patients were not. Furthermore, to study the relationship between changes in plasma VEGF levels by nitroglycerin treatment and response to DCb, 29 patients with advanced lung adenocarcinoma were treated with nitroglycerin for 3 days before chemotherapy using DCb.

Results: The rates of immunoreactive cells for HIF-1α, VEGF, and P-gp in tumor tissues treated with nitroglycerin were lower than those without nitroglycerin, but those for p53 were not different between those treated with and without nitroglycerin. Furthermore, the rates of immunoreactive cells for VEGF and P-gp proteins were significantly associated with those for HIF-1α in tumor tissue. The magnitude of decrease in plasma VEGF levels after treatment with nitroglycerin was significantly associated with response to DCb in patients with advanced lung adenocarcinoma.

Conclusions: Nitroglycerin treatment may improve response to DCb in patients with lung adenocarcinoma, partly through decreasing VEGF and P-gp production via reduction of HIF-1α.

Low levels of oxygenation due to relative vascular insufficiency have been shown to exist in solid cancers but not in normal tissues (1, 2), and hypoxia in solid cancers promotes stabilization of hypoxia-inducible factor 1α (HIF-1α; ref. 3). Hypoxic conditions in solid cancers are associated with a resistance to cancer therapy (4–6). HIF-1α activates the transcription of many genes that code for proteins involved in angiogenesis, cell growth, metastasis, and chemoresistance (7–13). Anticancer therapy to inhibit HIF-1α (11, 14, 15) and vascular endothelial growth factor (VEGF) have been reported (16). Furthermore, overexpression of P-glycoprotein (P-gp) in cancer tissue has been shown to efflux the intracellular anticancer drugs to the outside of cancer cells and to promote chemoresistance to taxanes including docetaxel (12, 13, 17).

The administration of nitric oxide (NO)–donating drugs decreased hypoxia-induced resistance to anticancer drugs in cancer cell lines (18) by means of reduction of HIF-1α (19–22) under hypoxic conditions, in vitro. Isosorbide dinitrate and inducible NO synthase gene transfer have various effects on tumor tissues and cells including augmentation of oxygen pressure in tumor tissue through an increase in blood flow (23), p53 protein activation, apoptosis, and growth inhibition in cancer cells (22, 24). In contrast, overexpression of inducible NO synthase promotes tumor angiogenesis and tumor progression (25).

A variety of anticancer drugs have been developed for the treatment of lung cancer and have contributed to prolonging survival (26). However, even third-generation regimens such as docetaxel and platinum result in response rates of 24% to 36% in the intention-to-treat population, and in 8 to 14 months in median overall survival among patients with advanced non–small cell lung cancer (NSCLC) with good performance status (27–29). We showed that a combinational use of nitroglycerin plus vinorelbine and cisplatin regimen improves the response rate to 72% and median time to progression (TTP) to 11 months in comparison with vinorelbine and cisplatin alone in previously untreated stage IIIB/IV patients with NSCLC in a phase II trial, showing the additional effect of nitroglycerin on vinorelbine and cisplatin regimen (30). Taxanes, including docetaxel and paclitaxel, are widely used for chemotherapy in patients with advanced NSCLC (29). However, the effects and their mechanisms of nitroglycerin on the increase in response to docetaxel and carboplatin regimen (DCb) in patients with lung adenocarcinoma, in vivo, have not been reported.

Patient characteristics. A total of 17 patients with operable lung adenocarcinoma were selected to investigate the long-term effects of nitroglycerin on HIF-1α, P-gp, VEGF, p53, and activated p53 (phosphorylated p53 at serine 15) proteins in their resected tumors by semiquantitative immunohistochemical analyses (study A). The 17 patients with lung adenocarcinoma and stable angina pectoris were operated on between January 2001 and February 2004 at the Department of Thoracic Surgery, Institute of Development, Aging, and Cancer, Tohoku University School of Medicine located in Miyagi Prefecture, Japan. Eight of the 17 patients (nitroglycerin group) had been continuously treated with nitroglycerin patches (25 mg/patient daily) before operation for at least 3 months. Nine of the 17 operable patients (control group) who had not been treated with nitroglycerin were selected as controls in this study. To exclude the effects of vasodilators such as calcium channel blockers on oxygenation of tumor tissues (31), patients treated with vasodilators were excluded from this study. Participant characteristics in study A are shown in Tables 1 and 2. Their age, gender, cancer staging, and tumor volume were frequency-matched between the nitroglycerin group and control group (Tables 1 and 2).

To study the effects of short-term nitroglycerin treatment for 3 days on the HIF-1 pathway including plasma VEGF levels (32) and on the relationship between change in plasma VEGF levels by treatment with nitroglycerin for 3 days and response to nitroglycerin plus DCb, a total of 29 patients with advanced lung adenocarcinoma were studied (study B). The 29 patients fit the following 15 inclusion criteria: (a) the diagnosis of lung adenocarcinoma was confirmed with histologic or cytologic examination, (b) age ≥40 years old, (c) no treatment with a vasodilator such as calcium channel blockers (31), (d) stage IIIB or stage IV disease, (e) no prior chemotherapy or radiotherapy, (f) a measurable or evaluable tumor lesion according to WHO criteria, (g) good performance status: a performance status of 0 to 1 according to the Eastern Cooperative Oncology Group scale (33), (h) without brain metastasis, (i) adequate hepatic function (serum bilirubin, alanine aminotransferase, and aspartate aminotransferase <2 × LN), (j) adequate hematologic function (neutrophil count >2,000/mL, hemoglobin >10 g/dL, platelet count >100,000/mL), (k) adequate cardiac function (cardiothoracic ratio <55%), (l) informed consent to receive chemotherapy and attend this study was obtained, (m) scheduled treatment with chemotherapy and without radiotherapy, (n) no ischemic heart disease, and (o) no enrollment in another interventional study. The 29 eligible patients with advanced lung adenocarcinoma treated with nitroglycerin patches (25 mg/patient daily for 5 days between 3 days before the start of each cycle of chemotherapy and cycle day 2) received chemotherapy using DCb (docetaxel, 60 mg/m² on day 1 and carboplatin target area under the curve (AUC) 5 on day 1; ref. 27) at the Department of Geriatric and Respiratory Medicine, Tohoku University School of Medicine located in Miyagi Prefecture, Japan, at the Department of Respiratory Medicine, Kesennuma City Hospital located in Miyagi Prefecture, Japan and at the Division of Internal Medicine, Furukawa City Hospital, located in Miyagi Prefecture, Japan between October 2002 and April 2006. The dosage of carboplatin used for chemotherapy was calculated by the Calvert formula (34). Participant characteristics are shown in Table 3.

To assess response to chemotherapy, we compared identifiable tumor sizes with a chest computed tomography scan before therapy and after the finish of the second and fourth cycles of chemotherapy. Response to chemotherapy was evaluated by two independent radiologists and an independent oncologist according to WHO criteria (35).

Peripheral venous blood was taken before and after treatment with nitroglycerin patches for 3 days before chemotherapy to measure plasma VEGF levels in the 29 patients. The change in plasma VEGF levels during nitroglycerin treatment for 3 days was calculated by subtracting plasma VEGF levels after nitroglycerin treatment from plasma VEGF levels before nitroglycerin treatment.

Staging of lung adenocarcinoma was determined using computed tomography scans of the brain, chest, and abdomen, positron emission tomography, gallium 67 citrate scintigraphy, and technetium 99m scintigraphy of the bone. Stage was defined using the revised lung cancer staging system of the American Joint Committee on Cancer (36). The histologic subtype of adenocarcinoma was classified according to the 2004 WHO classification (37).

Nitroglycerin transdermal patches (5-25 mg/patient daily) are widely and safely used in the treatment of coronary artery disease and heart failure (38). Therefore, we used 25 mg/body nitroglycerin transdermal patches daily as a NO donor. Nonsteroidal anti-inflammatory drugs were used if nitroglycerin-induced headaches occurred.

Estimation of the rates of immunoreactivity for HIF-1α, VEGF, P-gp, p53, and activated p53 proteins in resected cancer tumor tissue by semiquantitative immunohistochemical analyses. To assess the effects of nitroglycerin on the rates of immunoreactivity for HIF-1α, VEGF, P-gp, p53, and activated p53 proteins in resected cancer tumor tissue from patients treated with or without nitroglycerin patches, immunohistochemistry studies were done by a blinded specialized pathologist. The primary antibodies used in this study were goat polyclonal antibody HIF-1α (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit polyclonal antibody activated p53 (Cell Signaling Technology, Danvers, MA), and mouse monoclonal antibody VEGF (JH121; Lab Vision, Fremont, CA), P-gp (JSB-1; Invitrogen, Carlsbad, CA), and p53 (B20.1; Biomedia, Foster City, CA). Dilution of the antibodies was 1:100 for HIF1-α, 1:100 for VEGF and activated p53, and 1:40 for p53. Antibody for P-gp was prediluted at an adjusted concentration before purchase.

All pathologic samples were immediately fixed in 10% buffered formalin (Wako, Japan) at room temperature for several days. Formalin-fixed, paraffin-embedded tissue was sectioned at 2 μm and mounted on silane-coated slides. After deparaffinization, sections were first treated with 0.3% dehydrogen peroxide in 100% methanol for 10 minutes at room temperature in order to reduce endogenous peroxidase activity. Antigen retrieval was done for VEGF and p53 by microwaving for 15 minutes in citric acid buffer (2 mmol/L citric acid and 9 mmol/L trisodium citrate dehydrate, pH 6.0). Blocking was accomplished with 10% normal goat serum for HIF-1α and 10% normal rabbit serum for other antibodies in PBS containing 1% bovine serum albumin (Sigma, St. Louis, MO) for 30 minutes at room temperature. The sections were subsequently incubated in primary antibodies overnight at 4°C. After washing in PBS, slides were then incubated with biotinylated anti-rabbit or anti-mouse IgG for 30 minutes at room temperature, and subsequently incubated with peroxidase-conjugated streptavidin for 30 minutes at room temperature, using the Histofine Kit (Nichirei, Tokyo, Japan). The horseradish peroxidase reaction was carried out using diaminobenzidine solution [0.01 mol/L 3.3′-diaminobenzidine in 0.05 mol/L Tris-HCl buffer (pH 7.6), and 0.006% hydrogen peroxide] and counterstained with hematoxylin. For positive control, human colonic adenocarcinoma was used in this study. Brown staining of the nuclei of cancer cells was considered as positive for aberrant p53 protein expression (39).

The percentage of immunoreactive cells for HIF-1α, VEGF, P-gp, p53, and activated p53 proteins was analyzed by an independent specialized pathologist counting the number of immunoreactive cells in all cancer cells in at least 5 × 200 fields (×20 objective with ×10 ocular, 0.785 mm² per field). To evaluate the effects of oxygenation due to locations in tumor tissues such as the center and peripheral areas and the effect of histologic subtype components of adenocarcinoma on the immunoreactivity for HIF-1α, VEGF, P-gp, p53, and activated p53 proteins, we classified the histologic subtype components and graded the rates of immunostaining for HIF-1α, VEGF, P-gp, p53, and activated p53 proteins at the center and at the peripheral area in tumor tissues. A blinded pathologist assessed the immunoreactivity semiquantitatively as (−) if the rate of positively stained cancer cells for the target proteins was <10%, as (+) if the rate of positively stained cancer cells was between 10% and 50%, and as (++) if the rate of positively stained cancer cells was >50% in the tumor sections. We judged the sample as negative immunostaining if the immunohistochemical assessment was (−) and positive if the immunohistochemistry assessment was either (+) or (++).

Estimation of TTP after surgery in study A. The 17 patients with operable lung adenocarcinoma were followed up to study the effect of continuous use of nitroglycerin on TTP after surgery. The patients were evaluated by physical examination every 4 weeks and by complete blood cell count, biochemical tests, and chest radiograph every 3 months. If necessary, computed tomography scans of the brain, chest, or abdomen were appropriately done to assess disease progression. Computed tomography scans were reviewed by two independent radiologists to confirm disease progression.

Measurements of plasma VEGF levels using ELISA in study B. To study nitroglycerin effects on the HIF-1 pathway, we measured plasma VEGF levels by the ELISA method (40). Plasma VEGF levels were measured using a Quantikine human VEGF immunoassay kit (R&D Systems, Minneapolis, MN) before and after treatment with nitroglycerin patches for 3 days by a blinded technician, as previously described (40).

Study design. The primary end point in study A was to compare the rates of immunoreactive cells for HIF-1α, P-gp, VEGF, p53, and activated p53 proteins in tumor tissues between the nitroglycerin group and control group using immunohistochemical assessment. A secondary end point in study A was to ascertain the relationship between the rate of immunoreactive cells for VEGF and that for HIF-1α and the relationship between rate of immunoreactive cells for VEGF and that for P-gp in tumor tissues. Furthermore, we studied TTP after surgery in study A.

On the other hand, the primary end point in study B was to evaluate the relationship between changes in plasma VEGF levels by treatment with nitroglycerin for 3 days and responses to nitroglycerin plus DCb regimen. This study was approved by the Tohoku University Ethics Committee and informed consent was obtained from each subject.

Statistical methods. For statistical analysis in study A, gender, performance status, smoking history, cancer cell type, cancer staging, and immunostaining of HIF-1α, VEGF, P-gp, p53, and activated p53 proteins in cancer tissues were compared using Pearson's χ² contingency table analysis (or Fisher's exact probability test whenever appropriate) between patients treated with and without nitroglycerin. To exclude the effect of a subtype of adenocarcinoma on immunoreactivity (41), we compared the rates of immunoreactivity for HIF-1α, VEGF, and P-gp in only a papillary subtype of adenocarcinoma between the nitroglycerin group and control group using Fisher's exact probability test. Furthermore, the relationship between VEGF and HIF-1α immunostaining, and the relationship between VEGF and P-gp immunostaining in 17 patients with operable lung adenocarcinoma were evaluated by Fisher's exact probability test. Age, smoking history (pack-year), and tumor volume between patients with and without nitroglycerin treatment were compared using the unpaired Student's t test. Factors associated with positive immunostaining for HIF-1α, VEGF, and P-gp in resected tumor tissue such as age, gender, cancer staging, smoking history, histologic subtype of adenocarcinoma, and nitroglycerin treatment were calculated with logistic regression analysis. Relative risks (RR) and 95% confidence intervals (CI) were calculated to assess the rate of immunoreactivity for HIF-1α, VEGF, and P-gp.

TTP in study A was defined as the time from date of operation to date of disease progression. The progression-free rates were estimated using the Kaplan-Meier product-limit method. P values indicated the significance of differences between the nitroglycerin group and control group by log-rank test.

In study B, a comparison of plasma VEGF protein levels before and after treatment with nitroglycerin for 3 days was done by paired Student's t test. A comparison of change in plasma VEGF protein levels between responder and nonresponder to DCb was done with an unpaired Student's t test. Factors associated with response to DCb such as age, gender, performance status, cancer staging, tumor volume, and change in plasma VEGF levels during nitroglycerin treatment for 3 days were calculated with logistic regression analysis. RR and 95% CI were calculated to assess response to DCb. The cutoff value in change in the plasma VEGF levels before and after nitroglycerin treatment for 3 days in advanced lung adenocarcinoma discriminating responders from nonresponders to DCb were evaluated using receiver operating characteristics (ROC) curve analysis with ROCKIT (Windows 95, version 0.9.1 BETA) according to the ROCKIT User's Guide.¹⁰

We calculated sensitivity and specificity, and estimated the AUC, SE of AUC 95% CI of AUC of the change in the plasma VEGF levels before and after nitroglycerin treatment for 3 days in advanced lung adenocarcinoma using ROC curve analyses (42). All the statistical analyses in this study were carried out using the StatView program (SAS Institution, Inc., Cary, NC). P < 0.05 was considered significant.

Patient characteristics. There were no statistically significant differences in baseline characteristics regarding age (mean ± SE, 71.5 ± 3.9 versus 69.9 ± 3.5 years old; P = 0.761), gender (male/female, 2/6 versus 4/5; P = 0.434), cancer staging (with nitroglycerin: five patients in stage IA, two patients in IB, one patient in IIA, without nitroglycerin: seven patients in IA, one patient in IB, and one patient in IIB; P = 0.455), tumor volume (29,654 ± 13,844 versus 14,266 ± 3,073 mm³; P = 0.269), and smoking history (pack-years, 13.2 ± 8.7 versus 20.9 ± 9.2; P = 0.555) between the nitroglycerin group and control group before operation in study A as shown in Table 1. In the nitroglycerin group, five of eight patients had a dominantly papillary subtype, and three of eight had a dominantly acinar subtype of adenocarcinoma. In the control group in study A, eight of nine patients had a dominantly papillary subtype and one of nine patients had a dominantly acinar subtype. Thirteen of 17 patients (76%) had heterogeneous subtype components including papillary and acinar components of adenocarcinoma in a tumor tissue.

The characteristics of 29 patients (mean ± SE, 71.1 ± 2.1 years; male/female, 17/10; 13 patients in stage IIIB, and 16 in stage IV) with advanced lung adenocarcinoma treated with nitroglycerin plus DCb in study B are described in Table 3.

Comparison of HIF-1α, VEGF, P-gp, p53, and activated p53 immunoreactive cells in resected tumor tissues by immunohistochemical analyses between the nitroglycerin group and control group in study A. Typical immunohistochemical staining pictures of lung adenocarcinoma in the nitroglycerin group (A, C, E, G, and I) or in the control group (B, D, F, H, and J) in study A are shown in Fig. 1A and B, HE staining; Fig. 1C and D, HIF-1α immunostaining in center area of tumor; Fig. 1E and F, HIF-1α immunostaining in peripheral area in tumor; Fig. 1G and H, VEGF immunostaining; and Fig. 1I and J, P-gp immunostaining.

The HIF-1α immunostaining–positive cell rate in the tumor center (Fig. 1D) was higher than that in the tumor periphery (Fig. 1F) in the control group. The rate of patients (five of eight; 63%) negative for HIF-1α protein immunostaining in the nitroglycerin group was significantly higher than that (one of nine; 11%) in the control group (odds ratio, 13.33; 95% CI, 1.06-167.34; P = 0.027; Table 2). In four of seven patients in the control group, whose immunostaining for HIF-1α was positive, the rates of HIF-1α immunostaining–positive cells in the center were higher than those in the peripheral area in the tumor (Table 2).

The rate (five of eight; 63%) of patients negative for VEGF protein immunostaining in cancer tissues in the nitroglycerin group was significantly higher than that (one of nine; 11%) in the control group (odds ratio, 13.33; 95% CI, 1.06-167.34; P = 0.027; Table 2).

The rate (six of eight; 75%) of patients negative for P-gp protein immunostaining in cancer tissues in the nitroglycerin group was significantly higher than that (one of nine; 11%) in the control group (odds ratio, 24.00; 95% CI, 1.74-331.62; P = 0.008; Table 2).

Fisher's exact probability test shows that nitroglycerin was strongly associated with the negative immunostaining for HIF-1α (P = 0.007), VEGF (P = 0.007), and P-gp (P = 0.032) in a papillary subtype of adenocarcinoma. The rates and intensities of immunoreactivity for HIF-1α and VEGF in an acinar subtype of adenocarcinoma were slightly higher than those in a papillary subtype, although there was no statistical difference in the rates of immunoreactivity for HIF-1α and VEGF between an acinar subtype and a papillary subtype.

The p53 proteins detected by p53 immunostaining in the cancer nuclei were thought to be aberrant p53 by an independent specialized pathologist in this study. The rate (four of eight; 50%) of patients negative for aberrant p53 protein immunostaining in cancer tissues in the nitroglycerin group was not different from that (four of nine; 44%) in the control group (odds ratio, 1.25; 95% CI, 0.18-8.45; P = 0.819; Table 2). We tried to detect activated p53 immunostaining with polyclonal antibody for phosphorylated p53 at serine 15, however, we could not recognize the immunoreactivity for activated p53 in the fixed tumor tissues.

As described in Table 2, the rates of VEGF and P-gp protein immunoreactive cells were closely related to that of HIF-1α protein immunoreactive cells in resected cancer tissues in the 17 patients (P < 0.0001). Likewise, the rate of VEGF protein immunoreactive cells was also closely associated with that of P-gp protein immunoreactive cells in the 17 patients (odds ratio, 22.50; 95% CI, 1.60-317.35; P = 0.035).

In study A, the nitroglycerin treatment before the operation was significantly associated with the rate of immunoreactivity for P-gp (RR, 0.03; 95% CI, 0.00-0.71; P = 0.031) in tumor tissue after adjustment for age, gender, cancer staging, and smoking history in logistic regression analysis. Furthermore, the use of nitroglycerin was significantly associated with the rates of immunoreactivity for HIF-1α (RR, 0.05; 95% CI, 0.00-0.91; P = 0.043) and VEGF (RR, 0.05; 95% CI, 0.002-0.91; P = 0.043) after adjustment for age and cancer staging in logistic regression analysis. However, the use of nitroglycerin was not associated with the rates of immunoreactivity for HIF-1α (RR, 0.004; 95% CI, 0.00-3.18; P = 0.104) and VEGF (RR, 0.004; 95% CI, 0.00-3.18; P = 0.104) after adjustment for age, gender, cancer staging, and smoking history in logistic regression analysis. Furthermore, the use of nitroglycerin was not associated with the rates of immunoreactivity for P-gp (RR, 0.013; 95% CI, 0.00-1.12; P = 0.056) after adjustment for age, gender, cancer staging, smoking history, and histologic subtype of adenocarcinoma in logistic regression analysis.

Estimation of TTPs after surgery in study A. We followed-up the 17 patients with operable lung adenocarcinoma who underwent surgery in study A. The median follow-up periods were 1,172 days (range, 103-2,045 days) in the nitroglycerin group and 639 days (151-1,621 days) in the control group. Three of eight patients in the nitroglycerin group and three of nine patients in the control group were lost to follow-up. Two of eight patients (cases 1 and 5) in the nitroglycerin group and one of nine patients in the control group experienced disease progression. TTPs were 894 days in case 1 and 515 days in case 5 in the nitroglycerin group, and 174 days in case 7 in the control group. Kaplan-Meier analysis showed that the progression-free rate in the nitroglycerin group was not different from that in the control group (P = 0.857).

Treatment delivery in study B. In study B, composed of 29 patients, 15 (52%) patients received four courses of chemotherapy, 4 (14%) patients received three courses of chemotherapy, 8 (28%) received two courses of chemotherapy, and 2 (7%) received only one course of chemotherapy. The mean number of chemotherapy courses was 3.10 in study B. Of the 29 patients, 6 (21%) received chemotherapy at the full prescribed dose, and 22 (76%) patients received chemotherapy at the 80% dose intensity. The mean dose of docetaxel used in study B was 83.8%. The mean dose of carboplatin in study B was also 83.8%.

Treatment toxicity in study B. Treatment toxicities in study B are described in Table 4. Grade 1 headache was observed in 9 of 29 (31%) patients and grade 2 headache was observed in 3 of 29 (10%) patients during treatment with nitroglycerin. However, there were no severe headaches of grade 3 or more in this study. Grade 1 hypotension was observed in 6 of 29 patients (21%), 1 (4%) with grade 2 hypotension, and 1 (4%) with grade 3 hypotension during treatment with nitroglycerin. There was a high rate of severe neutropenia equal to or more than grade 3 and 4 in 26 of 29 patients (90%).

Plasma VEGF levels and response to DCb regimen in study B. In study B, in patients treated with nitroglycerin, following treatment with nitroglycerin patches for 3 days, plasma VEGF levels were significantly lower than the levels before treatment (mean ± SE, 218 ± 23 versus 181 ± 25 pg/mL; n = 29; P = 0.041; Fig. 2A).

Of 29 patients, 17 patients (59%) achieved partial response, 8 patients (28%) had stable disease, and 4 patients (14%) had progressive disease. The response rate to DCb was 59% (17 of 29 patients). Plasma VEGF levels before treatment with nitroglycerin in responders (248 ± 29 pg/mL, n = 17) to DCb regimen were not different from those in nonresponders (176 ± 36 pg/mL, n = 12; P = 0.127). The magnitude of decreases in plasma VEGF levels before and after nitroglycerin treatments for 3 days in responders to DCb (−80 ± 19 pg/mL, n = 17) were significantly larger than those in nonresponders (24 ± 23 pg/mL, n = 12; P = 0.002; Fig. 2B).

The change in plasma VEGF levels by nitroglycerin treatment for 3 days before chemotherapy (RR, 1.07; 95% CI, 1.01-1.13; P = 0.029) was significantly associated with response to DCb in advanced lung adenocarcinoma after adjustment for age, gender, performance status, cancer staging, and tumor volume in logistic regression analyses.

ROC curve analysis for change in plasma VEGF levels after nitroglycerin treatments for 3 days in study B. For the distinction between nonresponder and responder to the chemotherapy using DCb in advanced lung adenocarcinoma, the cutoff value of −16 pg/mL of the decrease in plasma VEGF levels after nitroglycerin treatments for 3 days yielded a sensitivity of 83.6% and a specificity of 83.5%. AUC of the ROC was 0.90, SE of the AUC was 0.07, and 95% CI of the AUC was from 0.71 to 0.98 (Fig. 2C).

To study the effects of nitroglycerin on the HIF-1 pathway and P-gp regulation in cancer cells in tumor tissues, we studied the rates of immunoreactive cells regarding HIF-1α, P-gp, and VEGF in the tissues by immunohistochemical assessment in 17 patients with operable lung adenocarcinoma and angina pectoris in study A. First, we compared the rates of immunoreactive cells for HIF-1α, P-gp, and VEGF proteins in tumor tissues between the nitroglycerin and control groups before operation. Furthermore, we studied the relationship between the rate of immunoreactive cells for VEGF protein and that for HIF-1α protein and the relationship between the rate of immunoreactive cells for VEGF protein and that for P-gp protein in cancer tissues. Second, a prospective trial was designed to evaluate the effects of nitroglycerin on the plasma VEGF levels which are regulated by HIF-1 (40), and the relationship between changes in plasma VEGF levels by nitroglycerin treatment for 3 days before chemotherapy and response to nitroglycerin plus DCb in 29 patients with stage IIIB/IV lung adenocarcinoma in study B. We showed that treatment with nitroglycerin dramatically decreased the rates of immunoreactive cells for HIF-1α protein in cancer tissues in patients with operable lung adenocarcinoma in study A. Furthermore, we showed that a decrease in plasma VEGF levels following nitroglycerin treatment for 3 days prior to chemotherapy was strongly associated with response to nitroglycerin plus DCb in patients with advanced lung adenocarcinoma.

Treatment with NO donors including nitroglycerin and sodium nitroprusside was reported to decrease HIF-1α in cancer cell lines under hypoxic conditions in a dose-dependent manner, in vitro (19–21). In an in vivo experiment, isosorbide dinitrate increased oxygenation in solid tumors through an increase in intratumor perfusion, not via regulation of tumor microvessels but via regulation of adjacent normal tissues and arterial blood pressure (23, 43). Furthermore, an increase in partial oxygen pressure reduces HIF-1α stabilization (19–21). Thus, findings in the previous studies suggest that the use of nitroglycerin might reduce the HIF-1α protein in solid tumors via an increase in intratumor perfusion and oxygenation in vivo, which are quite different from the circumstances in vitro. This is the first study demonstrating that treatment with nitroglycerin dramatically reduced the accumulation of HIF-1α proteins in solid tumor tissues in patients with operable lung adenocarcinoma.

In the present study, nitroglycerin treatment significantly reduced the rates of HIF-1α and VEGF immunostaining–positive cells in tumor tissues in operable lung adenocarcinoma as well as plasma levels of VEGF in advanced lung adenocarcinoma. Furthermore, the rate of immunoreactive cells for VEGF proteins were closely related to those for HIF-1α protein in the tumor tissue. The accumulation of HIF-1α under hypoxic conditions (44) in solid tumors activates the transcription of VEGF genes that code for proteins involved in angiogenesis (32, 39, 45). On the other hand, tumor growth depends on angiogenesis (46, 47). These events suggest that a decrease in HIF-1α proteins by nitroglycerin treatment may result in a decrease in the production and release of VEGF proteins from cancer tissues, and a decrease in angiogenesis in tumor tissues. However, Kaplan-Meier analysis in study A showed that continuous nitroglycerin treatment without chemotherapy did not have a clinical benefit in the prolongation of TTP after surgery in operable lung adenocarcinoma. We could not estimate the long-term effects of nitroglycerin on tumor growth and metastasis in advanced lung adenocarcinoma without chemotherapy. Further in vivo experiments and studies are needed to clarify the effects of nitroglycerin treatment on tumor growth and metastasis via angiogenesis.

We showed that a decrease in plasma VEGF levels after nitroglycerin treatment for 3 days before chemotherapy was strongly associated with response to nitroglycerin plus DCb in advanced lung adenocarcinoma. We also showed that the rate of immunoreactive cells for VEGF protein was closely associated with that for HIF-1α and P-gp proteins in cancer tissues. Nitroglycerin treatment before the operation was significantly associated with a reduction of immunoreactivity for P-gp in resected tumor tissue in logistic regression analysis. Furthermore, a decrease in plasma VEGF levels during nitroglycerin treatment for 3 days showed high sensitivity and specificity for response to nitroglycerin plus DCb in advanced lung adenocarcinoma. VEGF (32, 39) and P-gp proteins in cultured multicellular tumor spheroids by the use of human cancer cell lines (9, 48) are regulated by HIF-1 protein. High plasma VEGF levels were reported to associate with high microvascular density in tumor tissues and poor prognosis in NSCLC (49). The reduction of plasma VEGF levels in NSCLCs has been thought of as a therapeutic strategy, as shown previously (50). On the other hand, overexpression of P-gp proteins in cancer tissue has been known to associate with chemoresistance to several kinds of anticancer drugs such as taxanes (17) and cisplatin (51) in patients with cancer. These events suggest that decreases in plasma VEGF levels following nitroglycerin treatment and during chemotherapy may be associated with a preferable response to DCb via a decrease in HIF-1α pathways such as P-gp (9, 48) in tumor tissue. Actually, in clinical cases, we cannot directly monitor the protein levels of HIF-1α and P-gp in tumor tissue during nitroglycerin treatment. Therefore, a decrease in plasma VEGF levels by nitroglycerin treatment for 3 days may be a good indicator of response to DCb in advanced lung adenocarcinoma.

Because the histologic subtype components of lung cancer was reported to influence the distribution and intensity of the immunohistochemical staining by a monoclonal antibody (41), we compared the rates of immunoreactivity for HIF-1α, VEGF, and P-gp in only a papillary subtype of adenocarcinoma between the nitroglycerin group and control group. Fisher's exact probability test showed that nitroglycerin treatment was strongly associated with negative immunostaining for HIF-1α, VEGF, and P-gp in the papillary subtype of adenocarcinoma. However, the use of nitroglycerin was not associated with the rates of immunoreactivity for HIF-1α and VEGF after additional adjustment for age, gender, cancer staging, smoking history, and histologic subtype of adenocarcinoma in logistic regression analyses. Too many variables for logistic regression analyses in the small sample size and collinearity between male gender and smoking history may result in no statistical significances in logistic regression analyses, although the effects of histologic subtype of adenocarcinoma on the staining pattern cannot be completely excluded. Further studies are needed to clarify the effect of nitroglycerin treatment on HIF-1α, VEGF, and P-gp in solid cancer tissues in a larger number of patients with lung adenocarcinoma.

In addition to the HIF-1α pathway, several mechanisms may be associated with the regulation of chemosensitivity by nitroglycerin treatment; a NO donor, S-nitroso-N-acetyl-d-penicillamine, and inducible NO synthase gene transfer were reported to promote p53 activation in several cells in vitro, and increase radiosensitivity in animal models in vivo (22, 24). Moreover, activated p53 promotes p53-related apoptosis of cancer cells and an increase in chemosensitivity to anticancer drugs (52, 53). Aberrant p53 expression is strongly positively correlated with VEGF mRNA and interleukin-8 mRNA in NSCLC surgical specimens (38). Some NO donors including S-nitroso-N-acetyl-d-penicillamine increase the doxorubicin accumulation in a doxorubicin-resistant cancer cell line via reduction of the doxorubicin efflux rate by tyrosine nitration in the multidrug resistance–associated protein 3 (54). These events suggest that nitroglycerin may also improve the response to DCb via non–HIF-1 pathways such as accumulation of activated p53 protein (24, 52, 53) and via drug delivery with an increase in tumor blood flow (23, 43). However, we could not study the activated p53 by immunohistochemical analysis or the effect of nitroglycerin on the drug delivery change in fixed tumor tissues in the current study. Therefore, it is still uncertain what mechanisms mainly contribute to an increase in response rate in patients with lung adenocarcinoma treated with nitroglycerin. Moreover, the sample number in this study is small. A larger study is needed to elucidate the effect of nitroglycerin on HIF-1α, VEGF, P-gp, and activated p53 proteins and drug delivery in tumor tissue in patients with NSCLC.

In this study, the response rate to DCb in stage IIIB/IV lung adenocarcinoma was 59%, which is higher than those (24-36%) in previous studies (27–29). Furthermore, the rate of progressive disease to 14% DCb in the current study was consistent with that in a previous study (28). The rate of major adverse effects regarding nitroglycerin treatment, such as headache (41%), was also consistent with a previous report (30). However, the rate of hypotension (27%) in the current study was higher than that in our study (30). The increase in the rate of hypotension in this study may be due to the older population (median age, 71 years) than that of the previous study (64 years). These findings suggest that nitroglycerin may enhance a chemosensitivity to DCb with acceptable toxicity in patients with advanced lung adenocarcinoma. However, the data in the current study are not those of either a phase II or III trial but in a preliminary trial. Therefore, a large-scale randomized trial comparing nitroglycerin plus DCb with DCb alone in untreated stage IIIB/IV NSCLC is needed to show the additional effect(s) of nitroglycerin on the increase in response rate to DCb.

In summary, this is the first report demonstrating that the administration of nitroglycerin may decrease VEGF and P-gp proteins in the cancer tissue, and may decrease plasma VEGF levels via dramatic decreases in HIF-1α protein through an increased tumor tissue blood perfusion in responding patients with lung adenocarcinoma to DCb. Therefore, nitroglycerin may improve chemosensitivity to DCb in some patients with lung adenocarcinoma.

We thank Grant Crittenden for correction of the English text of the manuscript.