Adenosquamous histology predicts a poor outcome for patients with advanced-stage, but not early-stage, cervical carcinoma

Abstract

BACKGROUND

The objective of this study was to compare survival between patients with adenocarcinoma and patients with adenosquamous carcinoma of the cervix.

METHODS

Patients who were diagnosed with invasive cervical carcinoma from 1988 to 1999 were identified from the Automated Central Tumor Registry for the United States Military Health Care System. Clinical data, including race, age at diagnosis, histology, tumor grade, disease stage, lymph node status, treatment modality, and survival, were collected. Survival analysis was performed with Kaplan–Meier survival curves and compared using the log-rank test.

RESULTS

A total of 273 women were identified, 185 women with a histologic diagnosis of adenocarcinoma (AC) and 88 women with a diagnosis of adenosquamous carcinoma (ASC). Among the women with ASC, only 5% had Grade 1 tumors, and 66% had Grade 3 tumors. By comparison, among the women with AC, 37% had Grade 1 tumors, and 26% had Grade 3 tumors (P < 0.001). There was no difference in the incidence of positive lymph nodes or in the number of patients who underwent radical hysterectomy as primary treatment between patients with ASC and patients with AC. More patients with ASC received radiation therapy (51% vs. 28%) or chemotherapy (29% vs. 12%) as treatment (P < 0.001). Patients who had tumors with ASC histology had a significantly decreased 5-year survival rate compared with patients who had tumors with AC histology (65% vs. 83%; P < 0.002). When patients with early-stage cervical carcinoma (International Federation of Gynecology and Obstetrics [FIGO] Stage I) were examined separately, there was no statistically significant difference in the 5-year survival rate (AC, 89%; ASC, 86%; P = 0.644). However, when patients with advanced-stage disease (FIGO Stages II–IV) were analyzed, ASC was associated with a significant decrease in median and overall survival (P = 0.01). When the results were analyzed by grade, patients who had tumors with ASC histology had a shorter survival compared with patients who had AC histology of any grade; however, this was a significant difference only for patients with Grade 1 tumors: The 5-year survival rate for patients with Grade 1 AC was 93%, compared with 50% for patients with Grade 1 ASC (P < 0.01).

CONCLUSIONS

ASC histology appears to be an independent predictor of poor outcome in women with cervical carcinoma compared with their counterparts who have pure AC. The significant decrease in survival was observed only in patients with advanced-stage cervical carcinoma. This decreased survival may be related mainly to the grade of ASC. Cancer 2003;97:2196–202. Published 2003 American Cancer Society.

DOI 10.1002/cncr.11371

Cervical carcinoma is the third most common gynecologic malignancy in the United States.1 In 2001, it is expected that this disease will have affected 13,000 women and caused 4100 deaths.1 The number of women with cervical adenocarcinoma (AC) has been increasing over the past decades. From 1973 to 1998, the age-adjusted incidence of cervical AC increased by 29% compared with a decrease of 42% in their counterparts with squamous cell carcinoma.2

The observed survival rate for patients with cervical AC generally has been poorer compared with the survival rate for patients with cervical squamous cell histologies.2 The independent significance of AC histology in cervical carcinoma, however, remains controversial and is a subject of debate in the literature. Recent studies of cervical AC have been conflicting, showing both a poor prognosis and minimal or no importance as a predictor of overall survival in patients diagnosed with cervical carcinoma.3, 4 In addition, some authors have suggested that only the adenosquamous histology of cervical AC carries a poor prognosis, whereas other authors discount the significance of this particular histology.4–9 The objective of this study was to investigate any potential difference in survival between patients with cervical AC and patients with adenosquamous histology and to determine whether any histology is a significant, independent predictor of survival.

MATERIALS AND METHODS

A retrospective cohort study was performed of patients with cervical carcinoma in the United States Military Health Care System. Records were reviewed for those abstracted between July 1988 and July 1999 in the automated central tumor registry for the United States Military Health Care System through an approved protocol request. The United States Military Health Care System takes care of active duty members, their dependents or family members, and retired members of active duty service. This information is entered into each military treatment facility's tumor registry based on a report completed by a staff gynecologic oncologist. The data are then forwarded to the central tumor registry. All patients with cervical carcinoma who were entered into the registry and were included in this study were treated in the Military Health Care System by a gynecologic oncologist, either alone or in combination with a medical or radiation oncologist. Patients who were not treated primarily at a military institution were excluded.

Patients who were identified with invasive cervical AC were extracted for further analysis. Only patients with the histologic diagnosis of endocervical type AC or adenosquamous carcinoma (ASC) were extracted. Other AC histologies, including intestinal, signet ring, endometrioid, clear cell, small cell, glassy cell, serous, or villoglandular, were excluded. The variables studied included race, age at diagnosis, histology, grade, International Federation of Gynecology and Obstetrics (FIGO) stage, treatment modality, lymph node status, and survival. Histology was determined at the time of diagnosis by a pathologist and was reviewed at a central tumor board conference. Survival analysis was performed with Kaplan–Meier survival curves, and log-rank tests were used to determine the significance of difference between survival curves. Multivariable analysis of factors that influenced survival used Cox proportional hazards regression analysis. Significant differences in distribution of clinical factors were determined by the Wilcoxon rank-sum test.

RESULTS

A total of 310 patients were identified with the diagnosis of cervical AC. Thirty-seven patients were excluded due to histology other than endocervical (mucinous) AC or ASC. These included nine patients with endometrioid carcinomas, eight patients with papillary carcinomas, six patients with mucinous carcinomas, six patients with small cell neuroendocrine carcinomas, four patients with clear cell carcinomas, three patients with signet ring carcinomas, and one patient with serous carcinoma. This left 273 patients for analysis: 185 patients with pure AC and 88 patients with ASC. There was no difference in the mean ages between patients with AC (44 years) and patients with ASC (43 years; P = 0.4) (Table 1). The majority of patients with AC (81%) presented with early-stage disease, whereas only 68% of patients with ASC presented with Stage I disease (Table 1). Two hundred three of 273 patients (74%) had grade of histology determined in pathologic analysis. Patients with ASC histology tended to have a greater percentage of poorly differentiated tumors. Of the ASC tumors, 66% of tumors were Grade 3, whereas only 26% of AC tumors were poorly differentiated (P < 0.001) (Table 1). Lymph nodes were evaluated in 172 patients (63%). There was no difference in the proportion in each histologic group that had an evaluation of lymph node status: 66% of patients with AC compared with 55% of patients with ASC (P = 0.08). Ten of those 172 patients (6%) had positive lymph nodes. There was no difference between the percentages of patients with AC or patients with ASC who had positive lymph nodes (5% vs. 10%, respectively; P = 0.17). There was no difference in the percentage of patients with Stage IB disease in each histologic group that underwent radical hysterectomy as primary treatment 973% of patients with AC and 70% of patients with ASC; P = 0.65). More patients with ASC received radiation therapy as treatment (43% vs. 28% of patients with AC; P = 0.007) (Table 2). In addition, more patients with ASC were treated with chemotherapy (29% vs. 12% of patients with AC; P = 0.00004) (Table 2). All patients who received primary chemotherapy had concomitant radiation therapy.

Table 1. Clinical Characteristics of Adenocarcinomas

Characteristic	No. of patients (%)		P value
	Adenocarcinoma	Adenosquamous carcinoma
NS: not significant.
Age (mean, yrs)	44.5	43.4	NS
Disease stage	185	88	<0.04
IB	148.0 (80)	60.0 (68.0)	—
IIA	3.0 (1.5)	3.0 (3.5)	—
IIB	13.0 (7.0)	16.0 (18.0)	—
IIIA	0.0 (0.0)	2.0 (2.5)	—
IIIB	13.0 (7.0)	5.0 (5.5)	—
IVA	1.0 (0.5)	0.0 (0.0)	—
IVB	7.0 (4.0)	2.0 (2.5)	—
Tumor grade	145	58	<0.001
1	53 (37)	3 (5)	—
2	54 (37)	17 (29)	—
3	38 (26)	38 (66)	—
No. of lymph nodes evaluated	123 (66)	49 (55)	0.08
Negative	117 (95)	44 (90)	NS
Positive	6 (5)	5 (10)	—

Table 2. Treatment Modality

Treatment	No. of patients (%)		P value
	Adenocarcinoma	Adenosquamous carcinoma
Radical hysterectomy (Stage IB)	104 (73)	44 (70)	0.65
Radiation therapy	63 (34)	45 (51)	<0.007
External beam and brachytherapy	53	38	—
Other	10	7	—
Chemotherapy
All	22 (12)	26 (29)	<0.0004
Primary	16	20	—
Single agent	5	3	—
Multiagent	11	17	—
Salvage	6	6	—

Patients with ASC histology had a significantly decreased 5-year survival compared with patients who had AC (65% vs. 83%, respectively; P < 0.002) (Fig. 1). In a subset analysis of disease stage, patients with ASC histology continued to have a lower survival than patients with AC for certain stages and grades (Table 3). When patients with early-stage disease (FIGO Stage I) were examined separately, there was no statistically significant difference in 5-year survival (89% for patients with AC vs. 86% for patients with ASC; P = 0.644) (Table 3). However, when patients with advanced-stage disease (FIGO Stages II–IV) were analyzed, ASC histology was associated with a significant decrease in median and overall survival (P = 0.01) (Table 3). When analyzed by tumor grade, patients who had ASC histology had a shorter survival compared with patients who had AC for all grades, although this was a significant difference only for patients with Grade 1 tumors. The 5 year survival rate for patients who had Grade 1 AC tumors was 93% compared with 50% for patients who had Grade 1 ASC tumors (P < 0.01) (Table 3).

Figure 1. Kaplan–Meier survival curves for patients with adenocarcinoma and patients with adenosquamous carcinoma.

Table 3. Survival

Measure	Adenocarcinoma		Adenosquamous carcinoma		P value
	Median survival (months)	Five-year survival rate (%)	Median survival (months)	Five-year survival rate (%)
NS: not significant.
Stage
I	126	89	117	86	NS
II	122	80	40	38	—
III	54	32	15	0	<0.01
IV	7	14	13	0	—
Grade
1	123	93	15	50	<0.01
2	126	76	117	79	NS
3	64	57	100	59	NS

In a multivariate analysis using Cox regression adjusted for age, stage, grade, race, and histology, only stage and grade remained independent predictors of survival (P < 0.001). There was only a slight difference in the distribution of stage between the two groups analyzed, as stated above. This was due to the difference in distribution of patients with Stage I disease and patients with Stage II disease (Table 1). Both groups were similar with respect to the distribution of patients who presented with FIGO Stage III and Stage IV disease (10% and 11%, respectively) However, there was a large and significant difference between the groups with respect to the distribution according to tumor grade (Table 1): Only 26% of patients with AC had poorly differentiated tumors, compared with 66% of patients with ASC. Of the patients with early-stage (Stage I) disease, only 22% of patients with AC had Grade 3 tumors, compared with 56% of patients with ASC. There was an increase in the percentage of patients with late-stage (FIGO Stage II–IV) disease who had poorly differentiated tumors of both histologies compared with patients who had Stage I disease. Of the patients with advanced-stage AC, 38% had Grade 3 tumors, whereas almost all patients with advanced-stage ASC (82%) had Grade 3 tumors.

DISCUSSION

The histologic significance of cervical ASC has been limited somewhat by its inclusion as a subtype of AC in many of clinicopathologic analyses that have examined prognosis and outcome of patients with AC of the cervix.3, 4, 6–20 Previously, the majority of studies simply examined AC versus squamous cell histology.5–8, 10, 11, 13, 17–19 It is noteworthy that, in comparisons of squamous cell histology and AC, authors often will not differentiate between subtypes of AC or less common histologic variants.3, 5–8, 17–19 With the controversy regarding the significance of histologic subtypes and the assertion that patients with ASC histology have a worse prognosis, it is essential to search for this distinction and to determine why it may occur. In Table 4, we have segregated several published studies in the literature of AC versus ASC cell histology if they were noted separately and combined AC histology if they were not separated specifically for the analysis (Table 4). Five-year survival rates were taken directly from the publications or were abstracted from published Kaplan–Meier curves in each report. If 5-year survival was not stated explicitly in the publication or if Kaplan–Meier curves were not available, then studies were not included. Most published studies, if segregated by histology and stage, support a decline in prognosis for patients with ASC histology who have advanced-stage disease.3, 4, 6–20 The reported 5-year survival rates for patients with combined AC histologies in patients with FIGO Stage I disease and for patients with early-stage ASC all approached 70–80%, although there is a trend toward decreased survival for patients with ASC (Table 4). However, when patients with advanced-stage cervical carcinoma are evaluated together, there is a significant decrease in survival for patients with ASC on a stage-for-stage basis reported in the literature. Five-year survival rates for women with ASC were 27%, 0%, and 0% for patients with Stage II disease, Stage III disease, and Stage IV disease, respectively, compared with 59%, 14%, and 11%, respectively, for patients with AC and 48%, 17%, and 4%, respectively for patients with combined AC histologies.3, 4, 6–20 Our findings support the cumulative findings in the literature that patients with advanced-stage (FIGO Stage II–IV) ASC have a worse prognosis. This poor prognosis may be related to the prevalence of higher grade lesions in ASC histology throughout all stages. Nonetheless, the prognosis for patients with ASC, regardless of grade of tumor differentiation, generally is poor, with an approximately 63% 5-year survival rate for patients with all tumor grades (Table 3).

Table 4. Survival of Patients with Adenocarcinomas in the Reported Series

Study	Year	No. of patients
		Total	Disease stage		Five-year survival by disease stage
			I	II–IV	I	II	III	IV
AC: adenocarcinoma; ASC: adenosquamous carcinoma.
AC
Kjorstad10	1977	102	47	55	83	34	14	0
Moberg et al.11	1986	211	88	123	84	50	9	29
Gallup et al.12	1985	30	22	8	83	100	0	0
Look et al.13	1996	168	168	—	87	—	—	—
Lea et al.3	1988	98	98	—	60	—	—	—
Goodman et al.15	1988	35	35	—	79	—	—	—
Harrison et al.9	1993	70	62	8	78	—	—	—
Alfsen et al.4	2001	438	273	165	80	31	—	—
Current study	2002	273	208	65	89	80	32	14
Average	—	—	—	—	80	59	14	11
Combined histology: AC and ASC
Saigo et al.16	1986	136	94	52	76	41	11	11
Kleine et al.17	1989	144	64	80	76	49	27	0
Eifel et al.6	1990	334	223	111	75	35	27	10
Ishikawa et al.18	1998	193	144	49	88	45	0	0
Lea et al.3	2001	75	—	75	—	30	0	0
Eifel et al.7	1995	229	229	—	72	—	—	—
Kjorstad and Bond8	1984	150	150	—	88	—	—	—
Goodman et al.15	1988	21	—	21	—	90	38	—
Eifel et al.19	1991	160	160	—	82	—	—	—
Average	—	—	—	—	80	48	17	4
ASC
Gallup et al.12	1985	30	22	8	26	0	0	0
Look et al.13	1996	168	168	—	75	—	—	—
Hopkins et al.14	1988	98	98	—	60	—	—	—
Goodman et al.15	1988	14	14	—	86	—	—	—
Yazigi et al.20	1990	29	29	—	85	—	—	—
Harrison et al.9	1993	70	62	8	78	—	—	—
Alfsen et al.4	2001	438	273	165	55	44	—	—
Current study	2002	273	208	65	86	38	0	0
Average	—	—	—	—	69	27	0	0

Kjorstad, in his analysis of standardization treatment for patients with early-stage cervical carcinoma, revealed that patients with AC had a more aggressive course.10 This was based mainly on the observation that, after patients received preoperative radiation therapy alone, 30% of radical hysterectomy specimens had residual AC present at the time of surgery, compared with 11% of squamous carcinoma specimens. However, when Kjorstad evaluated the overall survival of patients with AC, he found no difference in the 5-year survival rate of 88% for patients with AC compared with patients from an earlier series who had squamous cell carcinoma.10

Eifel et al., in their multivariate analysis of patients with Stage IB carcinoma, found that AC histology was an independent clinical factor that worsened prognosis.7 This poor prognosis was demonstrated by the decrease in disease specific survival and central tumor control. That study compared squamous histology with AC histology. However, those authors did not segregate mucinous or endocervical AC from ASC histology. Consequently, no subset analysis was performed for comparison of ASC alone with AC or squamous cell histologies. Also, those authors sought to limit the analysis to tumors measuring > 4 cm in greatest dimension, again supporting the possibility that tumor size and (potentially) stage affects the prognosis of patients with ASC histology. In fact, when the authors evaluated patients with early Stage I AC, they found an 82% 5-year survival rate, which was comparable to the rate for patients with squamous histologies quoted in the literature, including their own.7, 19

Kleine et al. reported a worsened prognosis across stages for patients with AC in general compared with their counterparts who had squamous cell histology.17 Again, however, 21% of AC tumors were of ASC histology, and the authors did not differentiate histologies in their survival analysis. Those authors found no difference in survival for patients with early-stage AC who were treated surgically. However, among patients who received radiation therapy, patients with AC had a significantly decreased survival compared with patients who had squamous histology (58% vs. 85%, respectively).17 Kleine et al. made no comment on the tumor size of Stage IB AC tumors.

Eifel et al., in an analysis of clinically similar patients, reported an 88% 5-year survival rate for patients with Stage IB1 tumors who were treated primarily with radiation therapy, whereas the survival rate decreased to 45% for patients with Stage IB2 AC.6 Saigo et al. confirmed the hypothesis that lesion size and stage are the significant clinical factors in decreased prognosis for patients with ASC histology.16 Those authors found that stage was the most important variable for prognosis among patients with early Stage I and II disease, whereas the survival of patients with ASC histology was not significantly less compared with the survival of patients who had mucinous histologies.16

Gallup et al. found a marked decrease in survival in patients with ASC histology. The survival for patients with ASC was 20% compared with 83% and 80%, respectively, for patients with squamous histology and their counterparts with AC.12 This decreased survival was significant at all stages of ASC. Grade, however, was not assessed as a variable for survival.

In contrast, the analysis by Hopkins et al. of early Stage I cervical AC subtypes detected no evidence of histologic subtype as an independent factor for prognosis.14 Harrison et al. also concluded that, for patients with early Stage IB or IIA cervical carcinoma, the ASC cell type does not carry a worse prognosis compared with either pure squamous carcinoma or AC.9 In addition, in an analysis of 417 patients with AC, Alfsen et al. found that neither grade nor histologic subtype (with the exception of small cell carcinoma) was an independent prognostic factor for patients with AC.4 This lack of significance for ASC histology also was present for patients with advanced-stage disease.4

In our analysis, we believe that the worsened prognosis with increasing stage for patients with ASC histology was related mainly to higher tumor grade. This trend was seen throughout all grades but was significant in tumors with well-differentiated histology.

An analysis by Lea et al. of patients with AC demonstrated decreased survival with increasing tumor grade, with Grade 2 and 3 tumors associated with a significant decline in overall survival.3 Eifel et al. also noted this decline in survival associated with poorly differentiated AC.7, 19 Five-year survival rates were 67% versus 85% (P = 0.0018) in patients with Stage IB1, Grade 3 tumors who had AC histology and squamous histology, respectively.19

In addition, Hopkins et al. found a significant decline in the cumulative 5-year survival rate for patients with Stage I AC based on grade of tumor. Survival was 80% in patients with Grade 1 lesions, 69% in patients with Grade 2 lesions, and 41% in patients with Grade 3 lesions.14 The poor effect of grade on survival remained significant on multivariate analysis.

There also is molecular evidence that AC is unique and that larger, more poorly differentiated tumors may overexpress certain oncogenes.21–23 Kihana et al. found overexpression of c-erbB-2 in 39% of patients with Stage II and III cervical AC compared with only 9% of patients with early-stage AC.21 Over-expression of c-erbB-2 was associated with a poorer prognosis among all patients with AC who were analyzed. Mandai et al. found that overexpression of c-erbB-2 was associated with both a high incidence of lymph node involvement and a poor prognosis among the patients with AC who were analyzed, although this was not seen in corresponding patients with squamous carcinomas.23 Those authors concluded that increased expression of c-erbB-2, along with reduced nm23-H1 expression, has prognostic significance in patients with AC of the uterine cervix, whereas these proteins may not be associated with the prognosis of patients with squamous cell carcinomas.

ASC histology poses a significant threat as the incidence of nonsquamous cervical carcinoma continues to increase. The difficulty with characterization of signs and symptoms of early lesions and the inability to recognize early lesions consistently on Pap smears make this a dangerous entity among cervical carcinomas. ASC histology predicts poor survival for patients with AC of the cervix. This decreased survival is significant only for patients with advanced-stage disease (FIGO Stage II–IV). Finally, even when ASC tumors are classified as well differentiated, patients who have ASC tumors still have a significantly decreased 5-year survival rate (50%) compared with their counterparts who have AC (93%). Because of this finding, consideration should be made for interpreting all grades of ASC as an aggressive histologic variant and for treating patients with advanced-stage ASC using multimodal therapies.