Inadequacies of the current American Joint Committee on cancer staging system for prostate cancer

Authors


Abstract

BACKGROUND

Two major objectives of the American Joint Committee on Cancer (AJCC) staging system are to ensure appropriate treatments for patients and to determine prognosis. AJCC stage for distant prostate cancer includes patients with regional lymph node involvement. In the current study, the authors assessed whether patients with lymph node involvement and patients with distant metastasis, as determined using the Surveillance, Epidemiology, and End Results (SEER) staging system, had similar treatment and survival duration and, thus, were grouped together appropriately in the AJCC system.

METHODS

In total, 4141 patients were selected from The University of Texas M. D. Anderson Cancer Center's Tumor Registry who initially had registered at the center between January 1, 1982, and December 31, 2001, with a diagnosis of prostate cancer; had received no treatment before presentation; and had received treatment at the center. Patients with unknown stage and patients with any other primary malignancies were excluded. Descriptive analyses of demographic and disease variables were performed. Using SEER stage groups, survival analyses and Cox proportional hazards regression analyses were performed.

RESULTS

Treatments differed between patients with lymph node involvement and patients with distant metastasis. The median survival was 134 months for patients with lymph node involvement and 42 months for patients with distant metastasis. When these 2 groups were combined, as in the AJCC scheme, the median survival was 86 months.

CONCLUSIONS

The treatment and median survival of patients with lymph node involvement differed substantially from those of patients with distant metastasis. The current AJCC scheme for prostate cancer appeared to be inappropriate when considering its purpose, and the authors concluded that it should be revised. Cancer 2006. © 2005 American Cancer Society.

Staging for various malignancies was introduced first in 1929 by the League of Nations World Health Organization. The first cancer to be staged was cervical carcinoma. Staging was created because a standardized, coded format was needed that could be understood universally and that could be used for comparison of patients and groups of patients. The various stage groupings then could be evaluated for prognostic significance and therapy.1

Staging describes disease by using increasing values to represent the degree of involvement or severity of disease. For cancer, the extent of disease is based on how far the tumor has spread from the organ or primary site. Variables considered for any staging system are the primary tumor site, tumor size, number of tumors, depth of invasion and extension to regional or distant tissues, involvement of regional lymph nodes, and distant metastasis.

The two major staging systems that are in use currently in cancer registries are the American Joint Committee on Cancer (AJCC) staging system (Table 1) and the Surveillance, Epidemiology, and End Results (SEER) summary staging system (Table 2). The AJCC staging system was developed by the AJCC in cooperation with the Tumor, Node, Metastasis (TNM) Committee of the International Union Against Cancer (UICC). The TNM classification scheme is based on the premise that carcinomas of the same anatomic site and histology share similar patterns of growth and extension. AJCC staging is a scheme that can be captured on a staging form using TNM classification system and is applied universally. In the TNM classification system, the anatomic extent of disease is indicated. Clinical TNM status is based on evidence acquired before primary treatment, and the clinical stage is used to guide the selection of primary therapy. The pathologic stage can be used to assess the need for adjuvant therapy and the prognosis of the patient. Because the significance of criteria for defining the anatomic extent of disease differs for tumors located at different anatomic sites and tumors of different histologic types, criteria for TNM status must be defined for tumors at each anatomic site to attain validity.2

Table 1. Definitions for the American Joint Committee on Cancer (AJCC) TNM Classification System
  1. PSA: prostate-specific antigen.

Primary tumor, clinical (T)
 TX: Primary tumor cannot be assessed
 T0: No evidence of primary tumor
 T1: Clinically unapparent tumor not palpable nor visible by imaging
  T1a: Tumor incidental histologic finding in ≤ 5% of tissue resected
  T1b: Tumor incidental histologic finding in > 5% of tissue resected
  T1c: Tumor identified by needle biopsy (e.g., because of elevated PSA)
 T2: Tumor confined within prostate
  T2a: Tumor involves 1 lobe
  T2b: Tumor involves both lobes
 T3: Tumor extends through the prostate capsule
  T3a: Extracapsular extension (unilateral or bilateral)
  T3b: Tumor invades seminal vesicle(s)
 T4: Tumor is fixed or invades adjacent structures other than seminal vesicles: bladder neck, external sphincter, rectum, levator muscles, and/or pelvic wall
Regional lymph nodes (N)
 NX: Regional lymph nodes cannot be assessed
 N0: No regional lymph node metastasis
 N1: Metastasis in regional lymph node or nodes
Distant metastasis (M)
 MX: Distant metastasis cannot be assessed
 M0: No distant metastasis
 M1: Distant metastasis
  M1a: Nonregional lymph node(s)
  M1b: Bone(s)
  M1c: Other site(s)
Table 2. Surveillance, Epidemiology, and End Results (SEER) General Summary Stage
  1. NOS: not otherwise specified.

In situ
Noninvasive
Localized
 Invasive carcinoma confined to prostatic capsule (intracapsular)
 Invasive of prostatic capsule
 Prostatic urethra involved
 Localized, NOS
Regional by direct extension only
 Periprostatic tissues
 Seminal vesicle(s)
 Through prostatic capsule, including “fixation”
 Rectovesical (Denonvillier) fascia
 Bladder
 Rectum
 Extraprostatic urethra (membranous urethra)
Regional lymph node(s) involved only
 Hypogastric
 Iliac (common, internal, external)
 Obturator
 Pelvic, NOS
 Periprostatic
 Sacral (lateral sacral, sacral promontory, presacral)
 Direct extension and regional lymph nodes
Distant
 Direct extension or metastasis
  Skeletal muscles: Levator ani
  Pelvic bone
  Pelvic wall
  Ureter
  Sigmoid colon
  Penis
  “Frozen pelvis”
 Distant lymph nodes
  Aortic lymph nodes (paraaortic, periaortic, lumbar)
  Inguinal lymph nodes
  Other distant lymph nodes
 Other distant involvement
  Bone
  Brain
  Liver
  Lung

In contrast, the SEER staging system was designed to describe the extent to which a malignancy has spread from its point of origin. Its broad stage categories are consistent across all cancers and facilitate epidemiologic measurement of cancer efforts. The SEER staging system is used commonly in cancer registries; in fact, summary staging is a required data element for facilities and central registries that participate in the National Program of Cancer Registries of the Centers for Disease Control and Prevention. Although SEER staging is used frequently in cancer registries, it is not always understood by physicians, who are more likely to understand AJCC TNM staging.3

Regardless of which staging scheme is used, accuracy is necessary for predicting the outcome of patients with cancer, determining the most appropriate treatment regimen for these patients, comparing the results of different treatments nationally and worldwide, and reporting incidence rates of various cancers by stage. For example, although serum prostate-specific antigen (PSA) values, PSA doubling times, and Gleason scores are important predictors of survival for patients with prostate cancer and are used to guide prostate carcinoma therapy, a universal staging scheme still is necessary for reporting incidence rates and for comparing the results of different treatments nationally and worldwide. However, the two most widely used staging schemes (SEER and AJCC) vary considerably (Table 3).

Table 3. Stage Grouping for Prostate Carcinoma
Description of extentSEER general summary stageAJCC stageAJCC stage elementsa
  • SEER: Surveillance, Epidemiology, and End Results Program of the National Cancer Institute; AJCC: American Joint Committee on Cancer; NOS: not otherwise specified.

  • a

    Based on the American Joint Committee on Cancer tumor, lymph node, metastases (TNM) classification system.

Invasive carcinoma confined to prostatic capsule (intracapsular); invasive of prostatic capsule; prostatic urethra involved; localized, NOSLocalizedStage IT1a,N0,M0
  Stage IIT1b,N0,M0
   T1c,N0,M0
   T1,N0,M0
   T2,N0,M0
Periprostatic tissues; seminal vesicle(s); through prostatic capsule, including “fixation”; rectovesical (Denonvillier) fascia; bladder; rectum; extraprostatic urethra (membranous urethra)Regional by direct extension onlyStage IIIT3,N0,M0
Hypogastric; iliac (common, internal, external); obturator; pelvic, NOS; periprostatic; sacral (lateral sacral, sacral promontory, presacral)Regional lymph node(s) involved onlyStage IVAny T,N1,M0
Direct extension and regional lymph nodesDirect extension and regional lymph nodesStage IVAny T,N1,M0
Direct extension or metastasis to skeletal muscles (levator ani, pelvic bone, pelvic wall, ureter, sigmoid colon, penis, “frozen pelvis”); distant lymph nodes (aortic [paraaortic, periaortic, lumbar], inguinal, other distant lymph nodes);other distant involvement (bone, liver, brain, lung)DistantStage IVAny T,Any N,M1

For the current study, we compared the treatment and the overall survival of patients with prostate cancer using the SEER staging system. We were particularly interested in whether patients with regional lymph node involvement and patients with distant metastasis—two groups that are combined into one category in the AJCC staging system—receive similar treatment and have a similar median survival.

MATERIALS AND METHODS

Patients were selected from The University of Texas M. D. Anderson Cancer Center's Tumor Registry who initially had registered at the center between January 1, 1982, and December 31, 2001, with a diagnosis of prostate cancer. Only patients who had received no treatment before presentation at the center and had received any treatment at our facility were included in the study population. Patients with unknown stage were excluded, along with patients who had other malignancies in addition to prostate cancer. The resulting study population consisted of 4141 patients.

The stages of disease for patients were obtained from the Tumor Registry, where the 1977 SEER Summary Staging Guide scheme had been used for patients who registered between January 1, 1982, and December 31, 2000, and the 2000 SEER Summary Staging Manual scheme had been used for patients who registered between January 1, 2000, and December 31, 2001. To create a separate stage group for patients with regional lymph nodes, we then recoded the SEER stages using SPSS software into four groups according to their disease status, as follows: localized disease, regional extension, regional lymph node involvement with local or regional extension, and distant disease. By using SEER staging, we were able to separate patients with regional lymph node involvement from patients with distant disease. In contrast, the AJCC staging scheme does not distinguish between these two groups of patients. Race, residence, age, stage, grade, histology, and treatment within the first 4 months of presentation to The University of Texas M. D. Anderson Cancer Center were analyzed using frequency distributions in SPSS software (SPSS, Inc., Chicago, IL). Survival analysis was performed using the Kaplan–Meier method within SPSS software. Cox proportional hazards regression analysis was used to compute the relative risk of dying at each stage. A value of P < 0.05 was considered statistically significant.

RESULTS

Of 4141 patients with prostate cancer, 81% were white, 10% were black, and 8% were Hispanic. Approximately 33% of the patients were from Harris County, 40% were from elsewhere in Texas, 22% were from elsewhere in the U.S., and 5% were from other countries. The median age of the study population was 64 years (range, 13–86 yrs). Overall, 63% of patients had localized disease, 23% had regional extension, 7% had regional lymph node involvement, and 7% had distant disease. The distribution of disease stage changed greatly over time. During the earliest period (1982–1986), 39% of patients had localized disease and 21% had distant metastasis, whereas during the most recent period (1997–2001), the respective values were 77% and 4% (Table 4). Grade was available only as International Classification of Disease-Oncology (ICD-O) grade. Tumors were well differentiated in 25% of patients, moderately differentiated in 32%, poorly differentiated in 18%, anaplastic in 5%, and of unknown grade in 19%. Nearly all patients (98%) had adenocarcinoma. Overall, 40.9% of patients underwent surgery at The University of Texas M. D. Anderson Cancer Center, 31.5% received radiation therapy, 9.9% received endocrine therapy, and the remainder received combination therapy (Table 5). To determine whether the frequency of treatment modalities was the same for patients with regional lymph node involvement and patients with distant metastasis, we cross-tabulated the SEER stage groups and the treatment modalities. Of the patients with regional lymph node involvement, 82.4% underwent surgery as part of their treatment, 65.9% received endocrine therapy as part of their treatment (including patients who received endocrine therapy in the treatment categories “surgery + other” and “other”), and 8.8% received endocrine therapy as their only treatment. In contrast, the corresponding values for patients with distant disease were 6.9%, 81.8%, and 56.1%, respectively.

Table 4. Year of Patient Registration by Surveillance, Epidemiology, and End Results (SEER) Stage Group
Registration periodNo. of patients (%)
Localized diseaseRegional extensionRegional lymph node involvementDistant diseaseTotal
1982-1986115 (39.0)74 (25.1)45 (15.3)61 (20.7)295 (100)
1987-1991316 (45.5)179 (25.8)108 (15.5)92 (13.2)695 (100)
1992-1996738 (58.5)392 (31.1)57 (4.5)74 (5.9)1261 (100)
1997-20011454 (76.9)297 (15.7)63 (3.3)76 (4.0)1890 (100)
Total2623 (63.3)942 (22.7)273 (6.6)303 (7.3)4141 (100)
Table 5. Patient Treatment by Surveillance, Epidemiology, and End Results (SEER) Stage Groups
Treatment modalityNo. of patients (%)
Localized diseaseRegional extensionRegional lymph node involvementDistant diseaseTotal
  • a

    Includes patients who underwent surgery and received radiation therapy, chemotherapy, endocrine therapy, and/or immunotherapy in any combination.

  • b

    Includes patients who received with radiation therapy, chemotherapy, endocrine therapy, and/or immunotherapy in any combination.

Surgery only1246 (47.5)383 (40.7)61 (22.3)4 (1.3)1694 (40.9)
Radiation therapy only977 (37.2)307 (32.6)6 (2.2)15 (5.0)1305 (31.5)
Endocrine therapy only130 (5.0)88 (9.3)24 (8.8)170 (56.1)412 (9.9)
Radiation and endocrine therapy149 (5.7)63 (6.7)7 (2.6)24 (7.9)243 (5.9)
Surgery and endocrine therapy40 (1.5)25 (2.7)110 (40.3)11 (3.6)186 (4.5)
Surgery and othera42 (1.6)39 (4.1)54 (19.8)6 (2.0)141 (3.4)
Otherb39 (1.5)37 (3.9)11 (4.0)73 (24.1)160 (3.9)
Total2623 (100.0)942 (100.0)273 (100.0)303 (100.0)4141 (100.0)

Patients were followed for a median of 58 months (range, 1–245 mos). Survival analysis of the 4 SEER stage groups showed that the median overall survival was 197 months for patients with localized disease, 176 months for patients with regional extension, 134 months for patients with regional lymph node involvement, and 42 months for patients with distant metastasis (Fig. 1). Each of the four median values differed significantly from each other (P < 0.05 for all comparisons). Regression analysis revealed that, compared with patients who had localized disease, the risk of dying was 2.5 times greater for patients who had regional lymph node involvement and 10.1 times greater for patients who had distant disease (Table 6). When these 2 groups were combined, as they are in AJCC Stage IV, the median survival was 86 months and the risk of dying was 5.0 times greater than that for patients who had localized disease.

Figure 1.

Overall survival for patients with prostate carcinoma by stage.

Table 6. Relative Risk of Death for Patients by SEER Stage Group
SEER stage groupRR95% CIP value
  1. SEER: Surveillance, Epidemiology, and End Results Program of the National Cancer Institute; RR: risk ratio; 95% CI: 95% confidence interval.

Localized disease1.00Referent 
Regional extension1.731.44–2.01< 0.001
Regional lymph node involvement2.491.99–3.11< 0.001
Distant disease10.108.43–12.10< 0.001
Regional lymph node involvement and distant disease5.034.27–5.91< 0.001

To determine whether the time of diagnosis was a factor, survival analyses comparing patients who had regional lymph node involvement with patients who had distant metastasis were performed over four periods (1982–1986, 1987–1991, 1992–1996, and 1997–2001). For the 3 most recent time periods, a statistically significant survival benefit was observed for patients who had regional lymph node involvement (P < 0.05) whereas, during the period from 1982 to 1986, the survival analysis results trended toward a statistically significant survival benefit for patients who had regional lymph node involvement (P = 0.06).

A Cox proportional multivariate analysis using the prognostic factors available from our tumor registry (disease stage, ICD-O grade [Grades 1–4]), period of treatment, and patient age) demonstrated a statistically significantly better survival for patients who had lymph node involvement compared with patients who had distant metastasis (P < 0.05).

To determine whether there were survival differences between a subset of patients with lymph node involvement or distant metastasis who received treatment with the same modality, a Kaplan–Meier survival analysis was performed on patients who received endocrine therapy only (Fig. 2). That analysis demonstrated a 3.1 greater relative risk of dying for patients who had distant metastatic disease compared with patients who had regional lymph node involvement only (P = 0.001).

Figure 2.

Overall survival for patients who received endocrine therapy.

DISCUSSION

In the current study, we demonstrated convincingly that the prognosis differs greatly for patients with prostate cancer who have regional lymph node involvement compared with the prognosis for those who have distant disease. Compared with patients who had localized disease, patients who had regional lymph node involvement were 2.5 times more likely to die, and patients who had distant metastasis were 10.1 times more likely to die. Our multivariate analysis confirmed the prognostic value of stage when controlling for period of treatment, patient age at diagnosis, and ICD-O grade. Although it may be argued that prognostic variables, such as Gleason grade and serum PSA levels, were not included in the current analysis, we believe strongly that this does not diminish our ability to demonstrate survival differences between patients with regional lymph node involvement and patients with distant metastatic prostate cancer. Furthermore, differences in survival between these groups cannot be explained by treatment differences because a survival benefit was demonstrated for the regional lymph node group in a subgroup analysis of patients who were treated with endocrine therapy alone.

When both groups were combined, as in the AJCC staging system, both the median survival duration (86 mos) and the relative risk of dying (5.0) became nonrepresentative of either patient group. Therefore, AJCC Stage IV encompasses patients who may differ substantially in terms of survival and relative risk of dying.

We also demonstrated that the treatments given to patients with regional lymph node involvement were very different from those given to patients with distant metastasis. Physicians at The University of Texas M. D. Anderson Cancer Center apply best treatment practices to patients with prostate cancer and have developed different treatment approaches for these two groups of patients. Specifically, patients with lymph node involvement were far more likely to receive a local therapy (either surgery or radiation therapy) as a component of their overall treatment than patients with distant metastasis, who often were treated with endocrine therapy alone. Currently, most prostate cancer clinicians do not utilize stage in isolation to determine therapy for patients with prostate cancer. Over the last decade, tables or nomograms have been developed that incorporate serum PSA, Gleason score, and the clinical stage to predict the risk of adverse pathology or disease recurrence.4, 5 Many clinicians use these tables and nomograms in making treatment recommendations. Despite the fact that stage is only one of several variables used by clinicians to determine the best treatment plan, it remains an important variable. In fact, the National Comprehensive Cancer Network prostate cancer treatment guidelines suggest hormone therapy alone for patients with distant metastasis and either hormone therapy with radiation therapy or hormone therapy alone for patients with lymph node metastasis.6 Therefore, AJCC Stage IV encompasses patients who are likely to receive quite different initial therapies.

The dramatic shift in stage distribution over time has been noted in other prostate cancer studies and can accounted for by the advances in the early diagnosis of prostate carcinoma through the use of PSA tests.7, 8 Because of widespread use of serum PSA screening, patients now are presenting with earlier stage disease. Therefore, the incidence of regional lymph node metastasis and distant metastasis at presentation also is decreasing.9–11

The effect of positive lymph node status on the survival of patients with prostate carcinoma warrants additional research. Recent studies suggest that some patients with positive lymph node status who were treated by combined radical prostatectomy and adjuvant hormone therapy may have a more favorable outcome.12–15 In addition, hormone therapy combined with radiation therapy may offer better survival than hormone therapy alone.16 Therefore, many patients with positive lymph nodes currently receive hormone therapy combined with local therapy (either surgery or radiation therapy).

These findings suggest that the current AJCC stage groupings for prostate cancer are inappropriate for their purposes, which include the development of treatment plans and the prediction of patient survival. We have demonstrated that treatment and survival for patients with regional lymph node involvement differ substantially from those patients with distant metastasis, yet these distinct groups are combined into AJCC Stage IV. In its current format, AJCC Stage IV does not represent a homogenous group and therefore should not be used by practitioners to guide treatment decisions or to counsel patients on prognosis. The AJCC staging system should be revised to reflect these important differences.

Ancillary