SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information

Previously, we and others showed that hypoxia-inducible factor-1α (HIF-1α) and transcriptionally upregulated Aurora-A were required for disease progression in several tumors. Here, we address the clinicopathologic value of Aurora-A and HIF-1α in locally advanced nasopharyngeal carcinoma (NPC). Aurora-A and HIF-1α expression was semiquantitatively evaluated by immunohistochemistry staining in 144 cases from a randomized controlled trial. Of these patients, 69 received neoadjuvant chemotherapy plus concurrent chemoradiotherapy, and acted as the training set, and 75 cases treated with neoadjuvant chemotherapy plus radiotherapy were used as the testing set to validate the prognostic effect of Aurora-A and HIF-1α. We found that Aurora-A and HIF-1α were highly expressed in NPC, but were deficient in normal adjacent epithelia. In the testing set, Aurora-A overexpression predicted a shortened 5-year overall survival (59.1% vs 82.5%, P = 0.024), progression-free survival (44.8% vs 79.8%, P = 0.004), and distant metastasis-free survival (43.0% vs 17.3%, P = 0.016). Multivariate regression analysis confirmed that Aurora-A was indeed an independent prognostic factor for death, recurrence, and distant metastasis both in the testing set and overall patients. Moreover, a positive correlation between Aurora-A and HIF-1α was detected (P = 0.037). Importantly, although HIF-1α did not show any prognostic effect for patient outcome, the subset with Aurora-A and HIF-1α co-overexpression had the poorest overall, progression-free, and distant metastasis-free survival (all P < 0.05). Our results confirmed that Aurora-A was an independent prognostic factor for NPC. Aurora-A combined with HIF-1α refined the risk definition of the patient subset, thus potentially directing locally advanced NPC patients for more selective therapy. (Cancer Sci, doi: 10.1111/j.1349-7006.2012.02332.x, 2012)

Nasopharyngeal carcinoma (NPC), an Epstein–Barr virus (EBV)-related malignancy, has a remarkable racial and geographical distribution in Southeast Asia, especially in the Cantonese region of southern China.[1, 2] Compared with the early stage subset, local failure and distant metastasis-related cancer mortality caused a lower cumulative survival probability in locally advanced NPC.[3] However, the TNM staging system has limited power in determining outcomes for individual patients.[4] Supported by recently developed prognostic biomarkers, such as EBV DNA,[5-7] EBV DNase-specific neutralizing antibody,[8] epidermal growth factor receptor (EGFR),[9, 10] and Beclin 1,[11] the patient risk definition was refined more accurately. For stage II patients, the 5-year survival rates for EBV DNA high and low subgroups were 90% and 63%, respectively, leading to an altered risk definition for early stage subgroups.[5] Epidermal growth factor receptor was overexpressed in 90% of head and neck squamous cell carcinoma (HNSCC),[9] and predicted an inferior patient outcome.[12] More importantly, EGFR mAbs cetuximab and nimotuzumab significantly improve overall survival (OS) in combination with chemotherapy or radiotherapy for locally advanced NPC and other type of HNSCC.[12-15] Thus, to identify more novel molecular markers that could not only predict the prognosis individually but also provide potential therapeutic targets, will be of great benefit to individualized treatment for locally advanced NPC.

Mitotic Aurora-A kinase belongs to a serine/threonine kinase family comprising two other members, Aurora-B and Aurora-C.[16-18] We and others have found that Aurora-A is essential in proper timing of mitotic entry and formation of bipolar spindles.[19, 20] Overexpression of Aurora-A caused aberrant centrosome amplification, multipolar spindle structure, and aneuploidy.[21-23] Moreover, the malignant phenotype of Aurora-A might be ascribed to its interaction with several key molecular regulators, including Ajuba, p53, hypoxia-inducible factor-1α (HIF-1α), and TPX2.[20] In hepatocellular carcinoma, Aurora-A was transcriptionally upregulated by HIF-1α through binding hypoxia responsive elements within the Aurora-A promoter,[24] and was closely correlated with high tumor grade and p53 mutation.[25]

Indeed, Aurora-A overexpression has been found in a variety of malignancies, not only in solid tumors but also in leukemia, and predicted an inferior patient outcome.[25-29] Upregulation of Aurora-A mRNA, for example, was correlated with the occurrence of regional lymph node metastasis for HNSCC.[30, 31] Conversely, inhibition of Aurora-A by its specific small molecule inhibitor VX-680 potently suppressed the laryngeal HepG2 cell AKT1/2 phosphorylation as well as migration capacity, and sensitized the cell to X-ray irradiation.[32] For esophageal squamous cell carcinoma cells, inhibition of Aurora-A suppressed tumor growth and sensitized the cells to docetaxel chemotherapy.[29] Moreover, our previous study showed that suppression of Aurora-A by VX-680 led to 46.0% tumor growth suppression,[33] suggesting Aurora-A might be a promising therapeutic molecular target for NPC and other types of HNSCC. However, the prognostic effect of Aurora-A has not been characterized in human NPC.

In the present study, we addressed the clinicopathologic features of Aurora-A in 144 locally advanced NPC retrieved from a randomized controlled trial (RCT). We found that Aurora-A was an independent prognostic factor for locally advanced NPC. Moreover, a positive correlationship between Aurora-A and HIF-1α was detected. Importantly, we found that the subgroup with both Aurora-A and HIF-1α overexpression developed the worsened OS and distant metastasis-free survival (DMFS) for locally advanced NPC, suggesting that hypoxia and Aurora-A may enhance cancer mortality by promoting distant metastasis.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information

Patients and eligibility

A total of 408 patients were enrolled in a previous phase III RCT, aimed at comparing the therapeutic effects of induction chemotherapy and radiotherapy (IC/RT) with induction chemotherapy plus concurrent chemoradiotherapy (IC/CRT), from August 2002 to April 2005.[34] Of these, 144 randomized patients (69 IC/CRT + 75 IC/RT) were retrieved for the present study. The baseline of patient clinicopathologic characteristics of these two cohorts is shown in Table 1 and Figure S1. A strict eligibility criteria protocol was used in that RCT.[34] The routine staging work-up consisted of a detailed clinical examination, fiberoptic nasopharyngoscopy, MRI of the entire neck from the base of the skull, abdominal sonography, chest radiography, a complete blood count, and a biochemical profile. The patient TNM stage was classified according to the 1992 NPC staging system of China.[4] New Drug Statistical Treatment 8.0 software (Anhui Provincial Center for Drug Clinical Evaluation, Wuhu, China) was used to generate a random number table for patient assignment. This study was approved by the Clinical Ethics Review Board at the Cancer Center of Sun Yat-sen University (Guangzhou, China), and written informed consent was obtained from all patients at their recruitment.

Table 1. Correlation between Aurora-A expression and clinicopathologic characteristics of patients with nasopharyngeal carcinoma, allocated to training and testing sets*
VariableTraining set (n = 69)Testing set (n = 75)
High expression (n = 29)Low expression (n = 40)P-valueHigh expression (n = 40)Low expression (n = 35)P-value
  1. *Fisher's exact test. Patients in the training set received neoadjuvant chemotherapy plus concurrent chemoradiotherapy. Patients in the testing set were treated with neoadjuvant chemotherapy plus radiotherapy. Hypoxia, high hypoxia-inducible factor-1α (HIF-1α) expression; Normoxia, low HIF-1α expression; WHO, World Health Organization.

Age (years)
≤43.022230.13219160.999
>43.07172119
Gender
Male22340.36729280.589
Female76117
Histology
WHO II150.389460.500
WHO III28353629
T stage
T1 + T2 + T319260.99921250.104
T410141910
N stage
N0 + N116190.62818170.819
N2 + N313212218
TNM stage
III16250.62321250.104
IVa13151910
Failure pattern
Local regional relapse360.254930.997
Distant metastases126176
Disease-related death118176
HIF-1α
Hypoxia14180.81121130.246
Normoxia15221922

Oncologic treatment

In the IC/RT subset, patients were given two cycles of floxuridine + carboplatin (floxuridine 750 mg/m2, days 1–5; carboplatin AUC = 6) chemotherapy and underwent radiotherapy thereafter at 1-week intervals. In the IC/CRT subgroup, 1 week after completion of two cycles of floxuridine + carboplatin (floxuridine 750 mg/m2, days 1–5; carboplatin AUC = 6), patients received radiotherapy and concurrent carboplatin (AUC = 6) chemotherapy on days 7, 28, and 49.

The traditional Co60 γ-ray or linear accelerator 6–8 MV photon-based 2-D technique was used for radiotherapy. The accumulated radiation dose of 68–72 Gy, with 2 Gy daily fractions, 5 days per week, was delivered to the primary tumor. For lymph node negative invaded necks, 50 Gy irradiation was given, compared with 60–70 Gy radiation for lymph node positive invaded necks.

Semiquantitative assessment of immunohistochemical (IHC) staining

Immunohistochemical staining was carried out to detect the expression levels of indicated molecules. The primary antibodies of this study were rabbit anti-Aurora-A antibody (1:200 dilution; Upstate Biotechnology, Lake Placid, NY, USA) and mouse anti-HIF-1α antibody (1:200 dilution; Millipore, Billerica, MA, USA). Moreover, a negative control was also used by replacing the specific primary antibody with non-immune serum immunoglobulins at 1:200 dilutions. The brown granules in cytoplasm or nuclei were considered as positive staining. The expression level of Aurora-A and HIF-1α was scored by assessing staining intensity and extent. We scored the staining intensity as: negative (score 0); bordering (score 1); weak (score 2); moderate (score 3); and strong (score 4). Staining extent was graded into five parts according to the percentage of elevated staining cells in the field: negative (score 0); 0–25% (score 1); 26–50% (score 2); 51–75% (score 3); and 76–100% (score 4). The staining intensity score multiply by the staining extent score to get the overall score. Two independent pathologists (X.J.F. and J.X.), blind to follow-up data, were responsible for IHC staining evaluation. A third pathologist arbitrated when any discrepancy arose between these two pathologists.

Western blot analysis

Tissue samples preserved in liquid nitrogen were ground for the Western blot assay. Protein concentration was determined using the Bradford method with BSA (Sigma-Aldrich, St. Louis, MO, USA) as the standard recommended protocol. Equal amounts of protein (50 μg) were run on a 12% SDS-polyacrylamide gel and transferred to nitrocellulose membrane (Bio-Rad Laboratories, Hercules, CA, USA). The membrane was blocked and incubated with primary mouse anti-GAPDH antibody (1:4000; Ambion, Austin, TX, USA), rabbit anti-Aurora-A antibody (1:1000 dilution; Upstate Biotechnology), and mouse anti-HIF-1α antibody (1:1000 dilution; Millipore).

Selection of cut-off score for each biomarker “positive” expression

The receiver operating characteristic (ROC) curve analysis was used to select the cut-off point for each variable.[35] Briefly, the sensitivity and specificity for the patient outcome being studied at each score was plotted to generate a ROC curve. The score localized closest to the point at both maximum sensitivity and specificity, the point (0.0, 1.0) on the curve, was fixed as the cut-off score that might be correctly classified as patient outcome. To facilitate ROC curve analysis, all of the variables were dichotomized.

Follow-up

The patients in that RCT were all followed up with strict protocol. After the completion of therapy, patients were followed up at 3-month intervals during the first 3 years and at 6-month intervals thereafter. Overall survival was defined as the time from diagnosis to the date of death, or when censored at the latest date if patients were still alive. Progression-free survival (PFS) was defined as the time from diagnosis to the date of local failure/distant metastasis, or the date of death, or when censored at the latest date. Local failure-free survival (LFFS) was defined as the time from diagnosis to the date of local regional failure, or the date of death, or when censored at the latest date. Distant metastasis-free survival was defined as the time from diagnosis to the date of distant metastases, or the date of death, or when censored at the latest date.

Statistical analysis

The multivariate Cox proportional hazards model was used to estimate the hazard ratios (HR) and 95% confidence intervals (CI) for patient outcome. The survival probabilities discrimination between patient subsets in OS, PFS, LFFS, and DMFS were determined by Kaplan–Meier analysis and log–rank tests. All P-values quoted were two-sided and P < 0.05 was considered statistically significant. Statistical analysis was carried out using spss version 17.0 (SPSS, Chicago, IL, USA).

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information

Patient features and IHC analysis

A total of 144 NPC patients, including 69 patients received IC/CRT in the training set and 75 cases given IC/RT in the testing set, were enrolled in this study. The median duration of OS for the training and testing sets was 70.9 and 68.8 months, respectively. The Kaplan–Meier analysis showed that these two therapeutic regimens (IC/CRT versus IC/RT) had a similar survival probability in the training and testing sets (Fig. S1, P = 0.947). To develop a reasonable cut-off score of each factor for further survival analysis, we subjected each IHC score to ROC curve analysis with respect to patient outcome in the training set. Specifically, the Aurora-A IHC cut-off scores for OS, PFS, LFFS, and DMFS were 7.0, 7.0, 5.0, and 7.0, respectively. We thus selected 7.0 (high expression >7.0 vs low expression ≤7.0) as the uniform cut-off point of Aurora-A for survival analysis. Moreover, the HIF-1α IHC cut-off scores for OS, PFS, LFFS, and DMFS were 5.0, 5.0, 5.0, and 7.0, respectively. Similarly, a score of 5.0 was used as the cut-off point to distinguish patients with high or low HIF-1α expression. The potential prognostic factors, including age, gender, histological style, TNM stage, Aurora-A, and HIF-1α level, are listed in Table 1.

Both Aurora-A and HIF-1α were highly expressed in NPC samples, particularly in the tumor invaded zone, compared with the low IHC staining in normal paired tissues (Fig. 1A,B). Consistently, Western blot analysis revealed similar findings in NPC tissue and normal epithelia (Fig. 1C). Aurora-A was overexpressed in 47.9% (69/144) of NPC and 7.6% paired normal tissues, compared with high HIF-1α expression in 45.8% (66/144) of NPC and 6.9% paired normal tissues.

image

Figure 1. Aurora-A and hypoxia-inducible factor-1α (HIF-1α) expression in nasopharyngeal epithelia and locally advanced carcinoma. (A) Aurora-A was overexpressed in the tumor zone, but had lower expression in normal adjacent epithelia (original magnification, ×50). Panels to the right and below show representative Aurora-A expression with enlarged view (original magnification, ×400). (B) HIF-1α was overexpressed in tumor tissue compared to normal nasopharynx epithelia (original magnification, ×50). Panels to the left and above show representative HIF-1α expression with enlarged view (original magnification, ×400). (C) Western blot analysis of Aurora-A and HIF-1α expression in nasopharynx epithelia (N) and tumor (T) that originated from three locally advanced nasopharyngeal carcinoma patients. Equal loading of protein was determined by GAPDH.

Download figure to PowerPoint

Correlation between Aurora-A and patient outcome

We next asked if the Aurora-A level could be used as an independent prognostic biomarker for locally advanced NPC. As shown in Figure 2(A), Aurora-A overexpression, classified by the training set fixed cut-off score, predicted an inferior 5-year OS than the low expression subset (59.1% vs 82.5%, P = 0.024) in the testing set. Moreover, an evident 5-year PFS disadvantage was also observed when comparing the Aurora-A high expression subset with the low expression subgroup (44.8% vs 79.8%, P = 0.004, Fig. 2B). Importantly, further survival analysis showed that Aurora-A overexpression contributed to an increased 5-year tumor distant metastasis rate (43.0% vs 17.3%, P = 0.016, Fig. 2D), as well as a marginally elevated 5-year local failure rate (25.8% vs 6.2%, P = 0.059, Fig. 2C), in the testing set. When the training and testing sets were combined, similar OS (60.1% vs 82.5%, P = 0.004, Fig. 2E), PFS (49.0% vs 78.1%, P < 0.001, Fig. 2F), and DMFS (57.8% vs 83.8%, P < 0.001, Fig. 2H) differences between Aurora-A high and low subgroups were also observed. However, the Aurora-A expression level failed to distinguish the LFFS difference in overall patients (87.4% vs 79.0%, P = 0.191, Fig. 2G).

image

Figure 2. Kaplan–Meier estimates of overall survival (OS), progression-free survival (PFS), local failure-free survival (LFFS), and distant metastasis-free survival (DMFS) stratified by Aurora-A expression in the testing set and in overall patients. In the testing set, higher Aurora-A expression was closely correlated with poor OS (A), PFS (B), as well as DMFS (D), and was marginal with LFFS (C). For overall patients, the subset with higher Aurora-A expression had an inferior OS (E), PFS (F), and DMFS (H). The Aurora-A expression level, however, failed to distinguish the LFFS difference in overall patients (G).

Download figure to PowerPoint

Correlation between HIF-1α and patient outcome

We then addressed the prognostic effect of HIF-1α in locally advanced NPC. Consistent with a previous study,[36] HIF-1α level related survival difference were not observed regarding to OS (testing set: 69.3% vs 70.6%, Figure 3A, and overall set: 75.4% vs 68.0%, Figure 3E), PFS (testing set: 58.8% vs 64.1%, Figure 3B), LFFS (testing set: 77.4% vs 88.9%, Figure 3C and overall set: 79.9% vs 88.2%, Figure 3G) and DMFS (testing set: 69.9% vs 69.2%, Figure 3D, and overall set: 66.3% vs 76.7%, Figure 3H) in testing set and overall patients (all P value > 0.05), though a marginal difference was revealed in PFS for overall set (57.3% vs 71.1%, P = 0.04, Figure 3F).

image

Figure 3. Kaplan–Meier estimates of overall survival (OS), progression-free survival (PFS), local failure-free survival (LFFS), and distant metastasis-free survival (DMFS) stratified by hypoxia-inducible factor-1α (HIF-1α) expression in the testing set and in overall patients. In the testing set, significant survival differences were not observed between HIF-1α high and low subsets regarding OS (A), PFS (B), LFFS (C), or DMFS (D). For overall patients, the subset with higher HIF-1α expression had an inferior PFS (F), whereas this prognostic effect was not shown in OS (E), LFFS (G), or DMFS (H).

Download figure to PowerPoint

HIF-1α–Aurora-A coexpression and survival analysis

Correlation analysis showed that HIF-1α and Aurora-A levels positively correlated with each other in locally advanced NPC (correlation coefficient 0.174, P = 0.037). Importantly, the subset with HIF-1α and Aurora-A co-overexpression was correlated with an inferior OS (59.4% vs 85.8%, P = 0.008, Fig. 4A), PFS (44.7% vs 83.4%, P < 0.001, Fig. 4B), and DMFS (55.8% vs 88.3%, P < 0.001, Fig. 4D), but not LFFS (75.4% vs 92.5%, Fig. 4C), compared to the low level subgroup.

image

Figure 4. Kaplan–Meier estimates of overall survival (OS), progression-free survival (PFS), local failure-free survival (LFFS), and distant metastasis-free survival (DMFS) according to tumor Aurora-A and hypoxia-inducible factor-1α (HIF-1α) expression profile in overall patients. The subgroup with Aurora-A and HIF-1α co-overexpression had a worsened OS (A), PFS (B), and DMFS (D) compared to the co-downregulated subset. The Aurora-A and HIF-1α profile, however, failed to distinguish the LFFS difference in overall patients (C).

Download figure to PowerPoint

Multivariate analysis

As shown in Table 2, Cox regression analysis indicated that Aurora-A was a negative prognostic factor to predict OS (P = 0.039; HR, 2.802), PFS (P = 0.012; HR, 2.957), and DMFS (P = 0.037; HR, 2.817) in the testing set. In overall patients (Table 3), Aurora-A overexpression was a poor predictor for OS (P = 0.018; HR, 2.212), PFS (P = 0.002; HR, 2.520), and DMFS (P = 0.002; HR, 2.914), but not for LFFS (P = 0.199; HR, 1.801). More importantly, co-overexpressed Aurora-A and HIF-1α evidently increased the risk of death (P = 0.041; HR, 2.781), disease progression (P = 0.001; HR, 4.261), and distant metastasis (P = 0.003; HR, 4.774), than Aurora-A alone for overall patients (Table 3). However, the significantly prognostic effect was not observed regarding to HIF-1α, gender, age, tumor, or node stage (Tables 2, 3).

Table 2. Multivariate Cox proportional hazards analysis in patients with locally advanced nasopharyngeal carcinoma treated with neoadjuvant chemotherapy plus radiotherapy (n = 75)
Variables (>cut-off point versus ≤cut-off point)Testing set
OSPFSLFFSDMFS
Hazard ratio (95% CI)P-valueHazard ratio (95% CI)P-valueHazard ratio (95% CI)P-valueHazard ratio (95% CI)P-value
  1. The Wald test was used to calculate P-values. CI, confidence interval; DMFS, distant metastasis-free survival; LFFS, local failure-free survival; OS, overall survival; PFS, progression-free survival.

Gender, male vs female1.725 (0.694–4.286)0.2411.136 (0.491–2.629)0.7660.406 (0.083–1.993)0.2671.639 (0.657–4.089)0.289
Age, >43.1 years vs <43.1 years1.234 (0.531–2.868)0.6260.880 (0.413–1.874)0.7400.421 (0.117–1.509)0.1840.938 (0.398–2.212)0.883
Tumor stage, >T4vs <T41.254 (0.506–3.106)0.6241.267 (0.564–2.845)0.5671.293 (0.352–4.755)0.6991.583 (0.633–3.955)0.326
Node stage, >N2vs <N20.903 (0.367–2.222)0.8241.136 (0.507–2.549)0.7571.344 (0.368–4.910)0.6541.111 (0.438–2.819)0.825
Aurora-A, >7.0 vs ≤7.02.802 (1.055–7.442)0.0392.975 (1.274–6.949)0.0123.344 (0.846–13.224)0.0852.817 (1.062–7.472)0.037
HIF-1α, >5.0 vs ≤5.00.810 (0.336–1.955)0.6390.952 (0.441–2.055)0.9001.519 (0.458–5.037)0.4940.832 (0.345–2.007)0.682
Table 3. Multivariate Cox proportional hazards analysis in overall patients with locally advanced nasopharyngeal carcinoma (n = 144)
Variables (>cut-off point versus ≤cut-off point)Overall set
OSPFSLFFSDMFS
Hazard ratio (95% CI)P-valueHazard ratio (95% CI)P-valueHazard ratio (95% CI)P-valueHazard ratio (95% CI)P-value
  1. *High hypoxia-inducible factor-1α (HIF-1α) (≥7.5) and Aurora-A (≥5.0) expression versus low HIF-1α (<7.5) and Aurora-A (<5.0) expression. The Wald test was used to calculate P-values. CI, confidence interval; DMFS, distant metastasis-free survival; LFFS, local failure-free survival; OS, overall survival; PFS, progression-free survival.

Gender, male vs female1.349 (0.677–2.687)0.3950.918 (0.478–1.765)0.7980.509 (0.146–1.770)0.2881.049 (0.511–2.154)0.897
Age, >43.1 years vs <43.1 years1.153 (0.627–2.121)0.6460.996 (0.577–1.720)0.9890.677 (0.277–1.657)0.3930.974 (0.523–1.812)0.933
Tumor stage, >T4vs <T41.349 (0.693–2.626)0.3791.130 (0.616–2.072)0.6931.192 (0.431–3.300)0.7351.155 (0.583–2.287)0.679
Node stage, >N2vs <N21.345 (0.697–2.596)0.3771.438 (0.793–2.605)0.2312.360 (0.839–6.642)0.1041.333 (0.681–2.610)0.402
Aurora-A, >7.0 vs ≤7.02.212 (1.143–4.281)0.0182.520 (1.399–4.539)0.0021.801 (0.734–4.417)0.1992.914 (1.459–5.817)0.002
HIF-1α, >5.0 vs ≤5.01.290 (0.693–2.401)0.4221.601 (0.918–2.790)0.0971.574 (0.657–3.770)0.3091.574 (0.836–2.965)0.160
HIF-1α and Aurora-A*2.781 (1.044–7.408)0.0414.261 (1.764–10.289)0.0012.776 (0.658–11.708)0.1644.774 (1.716–13.282)0.003

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information

Aberrant expression of Aurora-A was associated with inferior prognoses for several tumors. Although we previous proved that Aurora-A might to be a novel therapeutic target for NPC,[33, 37] its prognostic value had not been defined. Here, we randomly selected 69 IC/RT patients as the training set and 75 IC/CRT patients as the testing set, that had comparable clinical features (Table 1) and survival probability (Fig. S1), for this biomarker-guided prognosis analysis. We found that Aurora-A overexpression predicted a poorer survival rate in OS, PFS, as well as DMFS, and elevated HIF-1α predicted an inferior PFS for locally advanced NPC (Figs 2, 3). Further correlation analysis showed that Aurora-A and HIF-1α were positively correlated with each other. Importantly, Aurora-A and HIF-1α coexpression worsened the OS, PFS, and DMFS for locally advanced NPC (Fig. 4). Moreover, our multivariate analysis showed that Aurora-A was indeed an independent prognostic biomarker (Tables 2, 3).

Aurora-A kinase is essential in ensuring centrosome segregation and spindle formation.[19, 20] In tumors, aberrant expression of Aurora-A was negatively correlated with tumor stage, and lymph node as well as distant metastasis.[30, 33] Given its importance in tumor initiation and progression, targeting of Aurora-A might to be a novel therapeutic selection for cancer treatment. Combined with docetaxel, Aurora-A inhibition suppressed 44.0% tumor growth in esophageal squamous cell carcinoma.[29] Indeed, Aurora-A inhibitor VX-680 (MK-0457) showed a powerful capacity for suppressing tumor growth not only in leukemia,[38, 39] but also in solid tumors (pancreas, prostate, and colon cancers).[40] In laryngeal squamous cell carcinoma, VX-680 effectively sensitized 20–60% of cells to X-ray irradiation and prevented 50% of cells from migrating to distant areas by attenuating AKT1/2 signaling.[32] In a recent phase I study, MK-0457 was well tolerated and 48.1% (13/27) of adult solid tumors achieved stable disease,[41] offering a promising molecular targeting agent in cancer treatment.

Although the EGFR mAb cetuximab has achieved survival benefits for patients with HNSCC,[12, 42, 43] little therapeutic improvement for those with NPC has been evident in the last two decades. Thus, the development of more EGFR-like biomarkers, which not only act as prognostic factors but also as therapeutic molecules, would lead to the improvement of individualized medicine for locally advanced NPC. Our present study showed that, as well as being a therapeutic target,[33, 37] Aurora-A was an independent prognostic factor for OS, PFS, as well as DMFS (Tables 2, 3).

In laryngeal squamous cell carcinoma, inhibition of Aurora-A lowered 50.0% cancer cell migration capacity and led to an inferior OS (59.1% vs 87.9%).[32] We previously reported that Aurora-A induced epithelial–mesenchymal transition and invasion was mediated by activating cofilin and MAPK signaling.[33, 44] Moreover, high Aurora-A mRNA levels were inversely associated with tumor stage and distant metastasis (P < 0.001) in oral, pharyngeal, and laryngeal squamous cell carcinomas.[30] Here, we proved that overexpression of Aurora-A predicted an inferior OS and DMFS, indicating that Aurora-A-induced survival disadvantage might be attributed to its distinctive capacity in enhancing cancer cell migration to distant areas for NPC and other types of HNSCC.

Hypoxia microenvironment-induced chemoradioresistance and malignant tumor phenotypes, including increased invasiveness and metastases and poorer survival, were shown to be transcriptionally regulated by HIF-1α.[45] For locally advanced NPC, a positive hypoxic profile, defined as high expression of both HIF-1α and CA9, predicted an inferior PFS (P = 0.04).[36] In another small cohort (78 cases) of advanced NPC, however, HIF-1α alone or coexpressed with COX2 failed to demonstrate any prognostic value.[46] Here, we further proved that HIF-1α was not an independent prognostic factor for locally advanced NPC (Tables 2, 3). In hepatoma, Aurora-A was transcriptionally upregulated by HIF-1α through binding the Aurora-A promoter hypoxia responsive elements.[24] However, compared to the prognostic effect of Aurora-A in OS, PFS, and DMFS, HIF-1α predicted a poorer survival rate only for PFS (Fig. 3F). For the underlying mechanism, we supposed that, besides being transcriptionally upregulated by HIF-1α, Aurora-A might promote disease progression by interacting with other important cellular regulators, such as p53, CENP-A, Ajuba, and TACC.[47] Thus, HIF-1α and Aurora-A, although positively correlated, might render the distinctive force (Fig. 4, Table 3) in managing disease progression, leading to varied function in determining patient outcome for NPC.

Taken together, our study indicated that Aurora-A was an independent prognostic factor in OS, PFS, and DMFS for locally advanced NPC. Aurora-A was positively correlated with HIF-1α. Aberrant activation of Aurora-A and HIF-1α promoted disease progression, and worsened outcomes for patients with advanced NPC.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information

This work was supported by the National Natural Science Foundation of China (Grant Nos. 30772476 and 30873084 to Q.L., No. 81000934 to X.B.W., and No. 81071890 to M.Y.C.), and the Fundamental Research Funds for the Central Universities (X.B.W.).

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
cas2332-sup-0001-FigureS1.jpgimage/jpg146KFig. S1. Kaplan–Meier estimates of (A) overall survival, (B) progression-free survival, (C) local failure-free survival, and (D) distant metastasis-free survival according to therapeutic regimens in overall patients. IC/CRT, induction chemotherapy plus concurrent chemoradiotherapy; IC/RT, induction chemotherapy plus radiotherapy.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.