Estimated relationship of p53 expression to relapse-free survival.
Lassaletta L, Brandáriz JA, Benito A, de la Cruz J, Gómez C, Ballestín C, Hitt R, Colomer R, Álvarez-Vicent JJ. p53 Expression in Locally Advanced Pharyngeal Squamous Cell Carcinoma. Arch Otolaryngol Head Neck Surg. 1999;125(12):1356-1359. doi:10.1001/archotol.125.12.1356
Prognosis for advanced head and neck squamous cell carcinoma remains poor despite advances in treatment, although a small number of patients may benefit from induction therapy leading to increased local control. Mutations of the p53 gene, which are present in a considerable percentage of head and neck squamous cell carcinomas, have been associated with poor response to chemotherapy. The role of p53 protein overexpression (which is associated with p53 gene mutations) in predicting the response to chemotherapeutic agents and survival rates is not clear.
To determine the association of p53 expression with chemotherapy response rates and disease-free survival rates in 62 patients with locally advanced pharyngeal cancer treated with induction cisplatin-5-fluorouracil chemotherapy between 1983 and 1995.
Historical cohort. Archival tissue from biopsies done before chemotherapy was immunohistochemically stained for the p53 tumor suppressor gene (clone D0-7; DAKO Corp, Glostrup, Denmark).
Positive staining for p53 occurred in 45 (73%) of 62 cases, with the percentage of reactive cells ranging from 35% to 98%. Chemotherapy response rates were higher in the p53-negative group (15/17 [88%]) compared with the p53-positive group (27/45 [60%]) (P=.07). The risk of recurrence was lower in the p53-negative group compared with the p53-positive group at 2, 3, and 5 years after treatment (P=.03, P=.01, and P=.007, respectively). The median relapse-free survival rates of patients in the p53-negative group was 16 months, whereas those with p53 protein expression demonstrated a median relapse-free survival time of 9 months (P=.07). In multivariate analyses, the only independent factor of relapse-free survival rates was age older than 70 years.
The present study shows a trend favoring p53 overexpression as a predictive and prognostic factor in locally advanced pharyngeal cancer treated with induction chemotherapy.
HEAD AND neck squamous cell carcinoma (HNSCC) is first diagnosed in most patients when it is at an advanced stage. Prognosis for patients with locally advanced tumors remains poor, despite advances in surgery and radiation therapy techniques, and combined modality therapy is usually recommended. The role of chemotherapy in HNSCC is being intensively reevaluated. Although randomized studies have failed to demonstrate a clear survival rate advantage with the addition of chemotherapy, patients who respond well seem to have a highly favorable prognosis.1 Identifying the factors associated with tumor response to chemotherapy and better survival rates would allow for selecting which patients with locally advanced HNSCC should receive this treatment.
Mutations in the p53 tumor suppressor gene are among the most common genetic abnormalities in human tumors, including HNSCC.2 Although several studies suggest that p53 mutations increase with the progression of HNSCC, the biological meaning of these events in the pathogenesis, progression, and vital prognosis of these neoplasms is not clear.3- 5 p53 Integrity is essential for normal cell growth and an efficient activation of apoptosis.6 Mutation of the p53 gene and the corresponding overexpression of the mutated p53 protein have been associated in vitro with resistance to antineoplastic agents.7 Thus, loss of p53 function may lead to chemotherapy resistance because of abrogation of p53-dependent apoptosis.
The aim of this study was to compare the association of p53 expression with chemotherapy response rates and disease-free survival rates in 79 patients with locally advanced pharyngeal cancer treated with induction cisplatin-5-fluorouracil chemotherapy. p53 Expression is a feasible indicator of p53 gene mutation.8
Patients chosen for this study had received induction chemotherapy (cisplatin plus 5-fluorouracil) for locally advanced pharyngeal cancer between 1983 and 1995. Eligibility criteria required that the patients have biopsy-proven stage III (n=17) or stage IV (n=45) squamous cell carcinoma of the oropharynx (n=31) or the hypopharynx (n=31), with no distant metastases and no prior treatment. Patients with cancer of the nasopharynx and those with any previous or simultaneous cancer of any location were excluded from this study. Those cases in which immunostaining could not be performed because of insufficient tissue or deterioration of the biopsy specimen (n=16) were also excluded. Patients (97% men [n=60], 3% women [n=2]) ranged in age from 40 to 74 years (median age, 57 years). The data collected on each patient included tobacco and alcohol use, performance status (Karnofsky scale), primary tumor site, TNM stage (T1-T4, N0-N3), American Joint Committee on Cancer stage (III-IV), histologic grade of tumor differentiation (well, moderately, or poorly differentiated), therapeutic strategy, and clinical course. Patients received 100 mg/m2 of intravenous cisplatin on day 1, plus 1000 mg/m2 per day of fluorouracil as a continuous intravenous infusion on days 1 through 5. Three courses were administered monthly with interruptions when necessary for resolution of signs and symptoms of toxic effects. Tumoral response was evaluated 3 weeks after the completion of chemotherapy. The responses of the primary tumor and the neck nodes were graded separately, and the lesser response was recorded as the overall response for the patient. Complete response was defined as total disappearance of all objective evidence of disease. Partial response required a 50% or greater decrease in the sum of the product of the diameters of each measurable lesion. Stable disease meant a less than 50% decrease in measurable tumor dimensions. After completion of chemotherapy, most patients underwent surgical resection (35 of 62) and/or radiation therapy (59 of 62).
Formaldehyde-fixed, paraffin-embedded tissues of pretreatment biopsy specimens from primary pharyngeal lesions were obtained to confirm the diagnosis of squamous cell carcinoma. Sections 4 to 5 µm thick were cut from each block and dried on capillary-gap glass slides (ChemMate Capillary Gap Microscope Slides; DAKO Corp) for immunohistochemical analysis. Tissue sections were deparaffinized and rehydrated in graded alcohols into water. Endogenous peroxide activity was quenched with methanol in hydrogen peroxide. After the tissue was placed in 10-mmol/L citrate buffer, it was boiled in cook pressure for 5 minutes.
Immunostaining was performed using the avidin-biotin-peroxidase method. After application of normal rabbit serum (1:10 dilution; DAKO Corp), the slides were incubated in an automatic machine Tech Mate 500 (DAKO Corp) with mouse monoclonal antibody anti-p53 (clone D0-7; DAKO Corp) at a 1:100 dilution for 25 minutes and a kit avidin-biotin-peroxidase complex optimized for the automatic procedure.
Counterstaining with hematoxylin for 1 minute was the last step. The slides were then dehydrated through graded alcohols into a xylene substitute and mounted with a coverslip using a standard medium. A formalin-fixed, paraffin-embedded sample of colon carcinoma immunoreactive to D0-7 antibody was processed as a positive control. Negative controls were obtained by excluding the primary antibodies or replacing them with nonimmune serum.
Immunostaining of the tumors for p53 was assessed quantitatively. The fraction of stained cells was determined separately by 2 pathologists (A.B. and C.B.) who were not aware of the case histories. A 35% nuclear reactivity cutoff point was established; ie, if less than 35% of tumor cells showed nuclear reactivity, the tumor was considered p53 negative; if 35% or more of tumor cells showed nuclear reactivity, it was considered p53 positive.
The Statistical Package for the Social Sciences Plus program9 was used for statistical analysis. The Fisher exact test and χ2 test10 were used to assess the statistical significance of the association of p53 expression with chemotherapy response and the clinicopathologic parameters. Stepwise logistic regression analysis was used to evaluate the association in the multivariate setting.11
The median follow-up of this study was 145 months, ranging from 36 to 424 months. First relapse and disease-free deaths were counted as events in the determination of failure-free survival, which was measured from the day chemotherapy began. Univariate disease-free survival analysis was estimated using the actuarial method, taking into account survival probability at the end of the first, second, third, and fifth years (95% confidence limits), and the Kaplan-Meier techniques.12
Comparison of survival probability at specific moments and the log-rank test were used to detect differences between curves.13,14 Multivariate survival analysis was performed by the Cox proportional hazard regression model.15 A P value of less than .05 was considered statistically significant.
Positive staining for p53 occurred in 45 cases (73%), with the percentage of reactive cells ranging from 35% to 98%. No p53 protein staining occurred in the adjacent normal cells, and the pattern of immunoreactivity was exclusively nuclear. The detection of nuclear p53 was significantly associated with alcohol intake (P=.03). Among the 45 patients with p53-positive staining, 44 (98%) were drinkers. No correlation between p53 expression and age, sex, tobacco use, localization, grade of differentiation, or TNM staging was found. No clinicopathologic parameters were significantly correlated with p53 expression in stepwise logistic regression analysis.
Complete responses occurred in 12 patients (19%) and partial responses in 29 (47%), for an overall response rate of 66%. Characteristics of neither the patient nor the tumor were significantly related to the probability of response to chemotherapy. There was a strong trend for the correlation of p53-negative results and better response to chemotherapy, although it did not reach statistical significance (P=.07) (Table 1). When stepwise logistic regression analysis was performed, no clinicopathologic parameters were significantly correlated with response to chemotherapy. Similar analysis showed a trend toward higher probability of response in the p53-negative group (P=.07).
The 5-year relapse-free survival rate for the entire group was 20%, with a median relapse-free survival time of 10.6 months. Analysis of relapse-free survival by patient age showed significant differences between patients older than 70 years and those 70 years and younger (P=.01). No significant differences were found with the other parameters (sex, performance status, tobacco and alcohol intake, site, grade of differentiation, or TNM staging). Relapse-free survival curves were calculated on the basis of p53 status. Table 2 gives relapse-free survival probabilities, which were significantly higher for the p53-negative than the p53-positive groups after 2-, 3-, and 5-year follow-up (P=.03, P=.01, and P=.007, respectively). The median relapse-free survival time of patients with tumors lacking p53 expression was 16 months, whereas those with p53 protein expression had a median relapse-free survival time of 9 months. Relapse-free survival analysis showed a trend toward higher risk of recurrence in the p53-positive group (log-rank test; P=.07) (Figure 1). In a multivariate analysis in which the variables introduced were p53 status, age, node status, and tumor site, only age greater than 70 years was an independent predictor of relapse-free survival (P=.03).
Although most patients with advanced HNSCC have a poor prognosis, a small subset of patients respond favorably to neoadjuvant chemotherapy. The aim of the present study was to determine whether p53 expression, which is related to the synthesis of mutated p53 protein, may be an indicator of response to therapy and clinical outcome in patients with locally advanced pharyngeal carcinoma.
Several studies on p53 expression suggest its value as a prognostic marker for HNSCC.16,17 It has been also reported that p53 mutations are associated with treatment resistance.18 Tumors showing p53 overexpression might be less sensitive to chemotherapy and more aggressive than those with functional p53. However, conflicting data result when the response to induction chemotherapy and the p53 status are correlated.
In a study by Dunphy et al,19 no correlation of p53 overexpression with response to paclitaxel and carboplatin chemotherapy was noted. Also, these authors found conversion of p53 protein from positive to negative in 42% of specimens after chemotherapy. It is of particular note that Lavielle et al20 demonstrated a statistical correlation between p53 expression and a complete response to chemotherapy. Furthermore, Bradford et al21 found a correlation between p53 expression and prediction of which patients were likely to achieve organ preservation with induction chemotherapy. Good response to chemotherapy occurred more frequently in the p53-positive group, although the result was not significant (P=.08). The basis for this discrepancy is unclear.
In agreement with our initial hypothesis, we found that the expression of p53 protein in the tumor cells provides both predictive and prognostic information. p53-Negative tumors demonstrated higher response rates to chemotherapy compared with tumors with p53 overexpression, the result leaning toward significance in both univariate and multivariate analyses. Moreover, p53 negativity was strongly predictive of a favorable relapse-free survival rate 2, 3, and 5 years after treatment, although this result was not confirmed by the multivariate analyses. p53 Expression was independent of other prognostic factors for pharyngeal cancer such as tumor stage.
Although there seems to be a correlation between p53 expression and response to chemotherapy, several factors need to be clarified: (1) In addition to p53-dependent apoptosis, there may be other, not-well-known modulators of tumor response to chemotherapy. Indeed, multiple resistance factors have been proposed for the development of resistance in HNSCC.22 (2) The effect of p53 in mediating cellular responses to DNA damage may be cell-type specific and/or lesion specific.23 (3) Immunohistochemical analysis of p53 expression seems to be a sensitive but not a very specific method for predicting p53 mutations.24
Although p53 protein expression is commonly observed in HNSCC, its prognostic value seems limited. Clinical factors are the most powerful predictors of tumor recurrence in patients with HNSCC. Unfortunately there does not seem to be a predictor of the efficacy of chemotherapy that is consistent across all studies. p53 Must not be taken solely as a prognostic marker, but also as a tumor marker that allows specific treatment to be undertaken targeting p53. It is now possible to design drugs that block cancer cell division, mimicking the p53 tumor suppressor function. On the other hand, for p53-positive tumors, restoring normal p53 function using gene therapy seems to be a promising approach for head and neck cancer therapy.
In conclusion, the present results are consistent with a role of p53 as a determinant of chemosensitivity in pharyngeal cancer. However, whether p53 genetic alterations are molecular phenomena associated with a worse response to chemotherapy and a poorer clinical outcome cannot be definitively inferred from the present study. Further investigations specifically focusing on this topic are in progress.
Accepted for publication July 14, 1999.
Presented at the First World Congress of Head and Neck Oncology, Madrid, Spain, December 3, 1998.
Corresponding author: Luis Lassaletta, MD, Servicio de O.R.L. Hospital Universitario La Paz, Paseo de la Castellana, 261, 28046 Madrid, Spain.