World Journal of Surgery

, Volume 30, Issue 5, pp 872–878

Extent of Disease at Presentation and Outcome for Adrenocortical Carcinoma: Have We Made Progress?

Authors

    • Department of SurgeryUniversity of California, San Francisco
    • UCSF Comprehensive Cancer CenterUCSF/Mount Zion Medical Center, University of California, San Francisco
  • Emily Reiff
    • Department of SurgeryUniversity of California, San Francisco
  • Quan-Yang Duh
    • Department of SurgeryUniversity of California, San Francisco
  • Orlo H. Clark
    • Department of SurgeryUniversity of California, San Francisco
    • UCSF Comprehensive Cancer CenterUCSF/Mount Zion Medical Center, University of California, San Francisco
  • Alex McMillan
    • UCSF Comprehensive Cancer CenterUCSF/Mount Zion Medical Center, University of California, San Francisco
Article

DOI: 10.1007/s00268-005-0329-x

Cite this article as:
Kebebew, E., Reiff, E., Duh, Q. et al. World J. Surg. (2006) 30: 872. doi:10.1007/s00268-005-0329-x

Abstract

Background

Adrenocortical carcinoma (ACC), a rare and aggressive malignancy, accounts for up to14% of adrenal incidentalomas. The only chance of cure for ACC is diagnosis at an early stage; therefore, a main indication for adrenalectomy in patients with adrenal incidentaloma has been the potential risk of ACC. Recent studies suggest that this has led to earlier stage of ACC at diagnosis, more curative operations, and better survival.

Methods

We analyzed data on ACC from The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database. Four equal time quartiles (1973–1979, 1980–1986, 1987–1993, and 1994–2000) were compared for changes in demographics, pathology, treatment, and cause-specific mortality.

Results

The average age was 51.2 years (range: 1–97), and 45.9% of patients were men. The average tumor size was 12 cm (range: 2–36 cm), and only 4.2% were ≤ 6 cm. Most (88%) patients had surgical resection of their tumor, and external beam radiotherapy was used in only 12% of patients. Between the time quartiles compared (as well as annually), there was no significant difference at presentation in age at diagnosis, sex, race/ethnicity, tumor size, tumor grade, the frequency of distant metastasis, and overall TNM stage. Low tumor grade, lower stage of ACC, later time quartile, and surgical resection were associated with a lower cause-specific mortality by univariate analysis (P ≤ 0.002) and by multivariate analysis (P ≤ 0.031).

Conclusions

Although adrenal incidentalomas have become a common indication for adrenalectomy, this has not resulted in patients with ACC being diagnosed earlier or treated at a lower stage of disease at the national level. The most important predictors of survival in these patients are tumor grade, tumor stage, and surgical resection.

Adrenocortical carcinoma (ACC) is a rare malignancy with a poor prognosis. Most patients present with an advanced stage of disease and have a median survival time of less than 12 months, even after complete tumor resection.1 Patients who have localized tumors and complete resections have longer survival times and a lower risk of ACC recurrence.14

About two thirds of all patients with ACC will have tumors that hypersecrete glucocorticoids and or androgens, and one third of patients will have classic Cushing’s syndrome. In addition to symptoms of hormonal excess, patients with ACC commonly present with an abdominal mass, weight loss, and other constitutional symptoms. A growing number of patients have asymptomatic and nonfunctioning ACC that is discovered incidentally on imaging studies done for none adrenal symptoms.5

Although there have been no major significant advances in the treatment of patients with ACC, some recent single-institution and national cooperative group studies from the United States and Europe have reported improved survival rates in contemporary cohorts.14,6 Some of these studies have also reported earlier stage at diagnosis and more curative operations being performed than in the past.3,4 The reasons for these findings are unclear, but some researchers have speculated that this may be due to the increased use of more sensitive imaging studies that incidentally identify ACC at an earlier stage.14,6 Indeed, ACC accounts for up to 14% of adrenal incidentalomas, depending on the tumor size and selection criteria used for adrenalectomy.5 Today, one of the most common indications for adrenalectomy is the risk of ACC in an adrenal incidentaloma.7

Since 1973, the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) program has collected demographic, clinical, pathologic, treatment, and survival data in patients with ACC. This public-use database has been used to study the clinical outcome of patients for a variety of cancers, and it offers the advantage of a large study cohort with a long follow-up time.811 The purpose of this study was to determine whether there have been changes in the demographics, clinical presentation, treatment, and outcome of patients with ACC in the United States.

METHODS

Study Design

A retrospective cohort design was used to analyze the SEER data from January 1, 1973, through December 31, 2000. The study was reviewed and approved by the University of California, San Francisco, Committee on Human Research.

Data Source and Setting

The SEER cancer registries collect data on approximately 26% of the United States population.12 The SEER 12 Registry Database consists of cases diagnosed from 1973 through 2000 in the geographic areas consisting of the states of Connecticut, New Mexico, Hawaii, Iowa, and Utah; the metropolitan areas of Atlanta, Detroit, San Francisco–Oakland, and Seattle–Puget Sound, and cases diagnosed from 1992 through 2000 in the Los Angeles and San Jose-Monterey metropolitan areas and the Alaska Native region. The SEER database includes information on demographics, pathology, treatment, and survival.

Variable Definitions

Adrenocortical carcinoma cases were selected from the SEER database for public use (SEER 12 Registry Database) and cases diagnosed at autopsy or by death certificate only were excluded from the analysis. The cutoff date for follow-up was December 31, 2000. Classification of stage of disease at diagnosis, tumor size, treatment, and race/ethnicity followed SEER definitions. Stage was defined as local (confined to the adrenal gland), regional (extension into adjacent tissue or lymph node involvement), and distant (metastatic). The SEER stages were also recoded into the TNM staging system.13 Four equal time quartiles (group I = 1973–1979, group II = 1980–1986, group III = 1987–1993, and group IV = 1994–2000) and annual proportions were compared for changes in patient demographics, extent of disease, treatment, and cause-specific mortality. The age-adjusted incidence of ACC and extent of disease were determined using the 2000 U.S. standard population.

Data Analysis

We used the SEER Stat 5.0.20 software to identify cases of ACC and to determine trends in the incidence of ACC and extent of disease. Adrenocortical carcinoma cases were selected by ICD-0-3 code for Site (C749) and Histology (8000, 8140, and 8370). Data are reported as mean ± standard deviation (SD) or number (percent). Student t-tests or analyses of variance for multiple group comparisons were used if normality could be assumed, otherwise Wilcoxon rank-sum, Kruskal-Wallis, or Spearman rank correlation tests were used. The Pearson χ2 test was used for categorical data. For univariate survival analysis, the log-rank test was used to compare the Kaplan-Meier events. For multivariate survival analysis, a forward, stepwise, Cox proportional hazards regression model was used.

The specific variables studied were age, sex, race/ethnicity, time quartile of diagnosis, grade of tumor, tumor size, extent of disease (local, regional and distant), TNM stage, surgical treatment, and radiation treatment. Patients who died from unrelated causes were censored in the survival analysis. The observed differences were assumed to be statistically significant if the probability of chance occurrence was less than 0.05.

RESULTS

A total of 725 cases of ACC were reported from 1973 to 2000 to the SEER program. The number of cases per year ranged from 9 to 46. The average age at diagnosis was 51.2 years, but patient age ranged from 1 year to 97 years. The overall male-to-female ratio was 1.0:1.2, with no significant difference between age groups. Most patients were white (87.3%) (Table 1). In 644 patients, ACC was the first malignant primary tumor diagnosed.
Table 1

Demographic, pathologic and treatment characteristics for 725 cases of adrenocortical carcinoma

Demographics

n (%)a

Age at diagnosis, years

51.2 ± 18.8

Women

392 (54.1)

Race

 

  White

633 (87.3)

  Black

50 (6.9)

  Other

42 (5.8)

  Era of diagnosis

 

    1973–1979

102 (14.1)

    1980–1986

145 (20.0)

    1987–1993

188 (25.9)

    1994–2000

290 (40.0)

Pathology

 

Diagnostic confirmation

 

  Positive histology

688 (94.9)

  Positive cytology

22 (3.0)

  Clinical

1 (0.1)

  Radiographic

6 (0.8)

  Unknown

8 (1.1)

  Tumor size, cm

12.0 ± 5.6

Histologic tumor gradeb

 

  I

32 (19.3)

  II

34 (20.5)

  III

57 (34.3)

  IV

43 (25.9)

SEER extent of disease

 

  Local

294 (40.6)

  Regional

130 (17.9)

  Distant

252 (34.8)

  Unstaged

49 (6.7)

TNM stagec

 

  I

16 (2.2)

  II

157 (21.7)

  III

130 (17.9)

  IV

252 (34.7)

  Unknown

170 (23.5)

Treatmentd

 

  Surgical resection

 

    Yes

555 (76.6)

    No

76 (10.5)

    Unknown

94 (12.9)

External beam radiation

 

   Yes

85 (11.7)

   No

628 (86.6)

   Unknown

12 (1.7)

a Plus-minus values are means ± SD.

b Tumor grade data was reported only for 23% of the cases. I = well differentiated, II = moderately differentiated. III = poorly differentiated, and IV = undifferentiated.

c Because tumor size data was collected starting in 1988, 170 cases of ACC were not recoded into the TNM staging system.

d The exact extent of surgical resection is not reported in the SEER database. Therefore, the effect of completeness and extent of surgical treatments were not analyzed as prognostic factors.

According to the SEER staging system, 40.6% of the patients had localized tumors (Table 1). A total of 76.5% of patients had resection of their primary tumor, and 11.7% of patients had external beam radiation therapy in addition to surgical resection. Unfortunately, no information on chemotherapy is collected in the SEER database. The diagnosis of ACC was confirmed by histology in most patients (94.8%). The average tumor size was 12 cm, and only 4.2% were ≤ 6 cm (Fig. 1). Patients with distant metastasis had larger primary tumors than patients with localized and regional tumors (mean: 13.8 cm versus 11.2 cm and 11.4 cm, respectively, P ≤ 0.004).
https://static-content.springer.com/image/art%3A10.1007%2Fs00268-005-0329-x/MediaObjects/268_2005_329_f1.jpg
Figure 1

Primary tumor size distribution in ACC. The average tumor size was 12 cm and the median was 11 cm.

The overall age-adjusted annual incidence for ACC was 0.72 per million in the United States and did not significantly increase from 1973 to 2000 (Fig. 2A). There was also no change in the incidence of localized ACC during the study period (Fig. 2B). There were no significant differences between the four time groups or annually in age at diagnosis, sex, race/ethnicity, tumor size, tumor grade, the frequency of local and regional disease, and overall TNM stage (Fig. 2C). There was a trend toward fewer cases of distant disease in the latter time groups (group I and II versus groups III and IV), but this was not significant (P = 0.110).
https://static-content.springer.com/image/art%3A10.1007%2Fs00268-005-0329-x/MediaObjects/268_2005_329_f2.jpg
Figure 2

Age-adjusted annual incidence of ACC over time. a. Overall incidence of ACC over 28 years. b. Incidence of localized, regional, and distant ACC over time. Incidence represents number of cases (y axis) per year (x axis) per one million persons in the 2000 United States population. c. Proportion of ACC extent of disease by time quartile.

The mean follow-up time was 41.8 months and the median was 15 months. The cause-specific mortality rate was 58.2% at 1 year and 88.4% at 5 years. Because the diagnosis of ACC may be inaccurate for localized tumors in the absence of locoregional invasion or distant metastasis, we compared the survival of patients with localized ACC to expected survival matched for age, sex, race, and year of diagnosis. The observed survival of patients with localized ACC was well below the expected survival (Fig. 3). Low tumor grade (grade I and II), lower stage of disease (local, and TNM stage I and II tumors), later time quartile at diagnosis (groups III and IV), and surgical treatment were associated with a lower cause-specific mortality by univariate analysis (P ≤ 0.002) and by multivariate analysis (P ≤ 0.031) (Table 2). Because date of diagnosis (time quartile) was evaluated as a predictor variable for survival, but it influences the censoring process for survival analysis, we also looked at 2-year and 5-year cause-specific mortality. Diagnosis with ACC at a later time (groups III and IV) still remained a significant predictor of a lower cause-specific mortality, even when adjusted for tumor grade, stage, and surgical treatment (P = 0.01).
https://static-content.springer.com/image/art%3A10.1007%2Fs00268-005-0329-x/MediaObjects/268_2005_329_f3.jpg
Figure 3

Observed and expected survival rates for localized ACC stage. The red line is the expected survival matched for age, sex, race, and year of diagnosis for localized ACC cases.

Table 2

Multivariate analysis of cause-specific mortality inadrenocortical carcinoma

 

Hazard ratio (± 95% CI)

P Value

Tumor grade

  

  I (reference)

1

 

  II

1.85 (0.74–4.62)

0.290

  III

2.92 (1.24–6.88)

0.014

  IV

2.94 (1.26–6.83)

0.012

Stage

  

  Localized (reference)

1

 

  Regional

1.55 (1.11–2.18)

0.013

  Distant

2.26 (1.69–3.02)

<0.001

Surgical resection

  

  No (reference)

1

 

  Yes

10.46 (0.44–0.95)

<0.001

Time quartile

  

  Group I (reference)

1

 

  Group II

10.83 (0.4–1.1)

0.230

  Group III

0.66 (0.44–0.95)

0.010

  Group IV

0.64 (0.40–0.85)

0.002

CI: confidence interval.

DISCUSSION

Contemporary studies of ACC suggest improved patient outcome, earlier diagnosis, and increased resectability of tumors.14,6 This finding, however, has been observed in single-institution and national cooperative group studies, which are subject to a selection bias and may not be reflective of the general population at large. We believe that our study, which used the largest sample and most comprehensive data available on cancer cases in the United States circumvents many of the limitations of previous studies. We found no significant change in the incidence of ACC in the United States over the 28-year time period studied, or in the extent of disease at presentation and the number of ACC that were treated surgically. However, we did find that the cause-specific mortality for ACC decreased with time.

The clinical characteristics of ACC in this study are similar to those reported in most smaller cohort studies.14,6,1416 There was a slight predominance of women, and the average age at diagnosis was 51 years, but we found no bimodal distribution in age at diagnosis. The average tumor size was 12 cm, similar to that reported in most other studies.14,6,1416 Although some investigators have suggested an increasing number of ACCs are diagnosed at a lower stage because of being discovered incidentally by imaging studies, we found no trend toward smaller primary tumor size and lower stage of disease at diagnosis over a 28-year period. Even though ACC accounts for up to 14% of adrenal incidentalomas, our study findings indicate that this has not necessarily resulted in earlier diagnosis of ACC at a lower stage at the national level.

The majority of ACCs in the SEER database were confirmed by histology. Pathologic findings such as a high number of mitoses, atypical mitoses, vascular or capsular invasion, and necrosis are typical histological features used to diagnose ACC.1 Several diagnostic histopathology scoring systems have been developed, but they are imperfect for distinguishing adrenocortical adenoma from carcinoma in the absence of obvious clinical locoregional invasion and or distant metastasis.1,15,17 Therefore it is possible that the patients with local ACC in this study could have been misdiagnosed. However, comparison of the survival rates of these patients to age, sex, and race-adjusted expected survival rates was significantly lower (Fig. 3). Furthermore, the tumor size for localized ACC was also larger than 6 cm in 93% of cases. Although we cannot be certain of the diagnosis of ACC for the localized tumors, these findings suggest that the localized tumors are more than likely to be malignant.

Our finding that survival rates were higher in patients diagnosed with ACC in the latter two time quartiles (groups III and IV) is similar to that of other recent studies.

Because there have been no significant advancements in the treatment of patients with ACC, it is unclear what the reason for this trend may be. Studies using mitotane therapy for adjuvant or primary treatment in patients with metastatic ACC have reported inconsistent response rates and are unlikely to account for an improved survival rate.1,16 We believe our study reflects a more accurate measure of patient outcome from ACC in the U.S., but is still limited because no data on chemotherapy and reoperations for recurrent ACC are available to determine their effect on survival.18 Because time of diagnosis was both a predictor variable and was used in the censoring process for survival analysis, it may account for the significantly different survival rates observed by era of diagnosis. However, when we addressed this issue by looking at 2-year and 5-year survival rates and the year of diagnosis as a continuous variable, era of diagnosis remained a significant predictor of survival.

The most important and consistently observed determinants of improved survival for patients with ACC have been localized tumors that are surgically resectable.14,6,1417,19,20 Some, but not all studies have found patient age, gender, primary tumor size, functional status of the tumor, tumor grade, histological score, and DNA ploidy are prognostic factors.14,6,1417,19,21 To our knowledge, our study is the largest cohort analyzed. We found that ACC stage, surgical resection, and tumor grade were the only independent prognostic factors for ACC. Further studies are necessary to better stratify those patients with localized ACC. Recent gene-profiling studies may be helpful for identifying molecular prognostic and diagnostic markers for ACC.22

Our study is limited by the fact that the SEER data represent approximately 26% of the US population and therefore do not permit an absolute measurement of the change in the extent of disease distribution for ACC. Nonetheless, the SEER database is the most comprehensive source of information on cancer in the U.S., and it is representative of the demographic and clinical characteristics of the U.S. population. Therefore, our inferences are based on the best population-based data available on cancer in the U.S. The SEER database is also subject to stringent quality control measures. Audits are undertaken to evaluate the accuracy and completeness of the data reported from participating cancer registries with 98% of the cases ascertained.12

In summary, despite some recent single-institution and national cooperative group studies from the United States and Europe that have reported earlier stage at diagnosis for ACC and more curative operations being performed than in the past, our study indicates that the extent of disease for ACC has not significantly changed at the national level in the United States. The most important predictors of survival in patients with ACC are tumor grade, tumor stage, and surgical treatment.

Copyright information

© Société Internationale de Chirurgie 2006