Age and Menopausal Status

Both patient characteristics have been extensively evaluated. When other, more important tumor characteristics are considered, age and menopausal status are not important prognostic indicators. A large study with long follow-up indicated that women 45 to 49 years of age had the best prognosis and that the very young (those under age 35 years) or elderly patients had the worst breast cancer survival.^{277, 278, 289, 312–315}

Axillary Lymph Node Involvement

Involvement of the ipsilateral axillary lymph nodes is still the most reliable and reproducible prognostic indicator for primary breast cancer (Table 121-9).^316–319 In general, 50 to 70% of patients with positive lymph nodes have a relapse, whereas only 20 to 45% of patients with all lymph nodes negative for metastatic disease have a relapse after locoregional treatments only. The risk of tumor recurrence in a patient with primary breast cancer is a continuum related to the number of positive axillary lymph nodes (see Table 121-9).³¹⁹ With each additional positive lymph node found, the risk of recurrence and metastasis increases by a few percentage points. Thus, patients with 4 to 10 positive lymph nodes have a greater risk than those with 1 to 3 positive nodes, and those with 10 or more positive nodes have a greater than 80% probability of recurrence and metastasis. Because nodal status cannot be accurately assessed by clinical means, an axillary lymph node dissection, including levels I and II, is considered the standard of care. Clinical studies have demonstrated that lymph node negativity is reliable only if at least 6, but preferably 10, axillary lymph nodes are removed and examined. Both macro- and micrometastasis within the lymph nodes have similar prognostic significance.^319–321 In recent years, primary breast cancer has been diagnosed in earlier and mostly localized stages. A classic axillary lymph node dissection has no therapeutic benefit for patients with node-negative axillae and is associated with considerable short- and long-term morbidity. An alternative (diagnostic) staging procedure for these patients is the sentinel lymph node biopsy.^{322, 323} This procedure limits considerably the extent of the surgical procedure in the axilla and, for the great majority of patients with negative axillary lymph nodes, precludes the need for a formal axillary dissection while providing similar (and in some cases superior) diagnostic and prognostic information. The identification of a single (or just a few) sentinel node also permits the pathologist to perform a more detailed assessment to detect micrometastases by combining light microscopy, IHC, and even more sensitive molecular techniques.³²⁴ The prognostic significance of identifying isolated metastatic cells in histologically negative lymph nodes by more sensitive techniques is still undetermined. Sentinel node biopsy and assessment are under extensive evaluation by large cooperative research groups, including the NSABP and the American College of Surgeons Oncology Group. The technique is making substantial inroads in the practice of breast surgery. Whether, and to what extent, it will displace or replace classic axillary dissection for early, localized breast cancer remains to be determined.

Table 121-9

5-Year Breast Cancer-Specific Mortality Rates According to Tumor Size and Axillary Lymph Node Involvement*.

Although axillary lymph node status is still the most powerful prognostic indicator, 15 to 45% of patients whose lymph nodes do not contain metastases still experience a recurrence and die. Conversely, up to 15% of patients with >10 positive lymph nodes treated only with surgery and radiotherapy survive without recurrence or metastases. Because of this limitation, other prognostic markers have been developed to improve prognostic accuracy, particularly in the group of patients with node-negative tumors.

Tumor Size

In addition to being a determinant for optimal local therapy, tumor size has prognostic significance in the determination of additional therapy (see Table 121-9). As the size of the tumor increases, the risk of recurrence or metastasis also increases, for both lymph node-negative and node-positive tumors. Since the risk of treatment failure is already high for patients with node-positive breast cancer, increasing tumor size adds relatively little prognostic value. However, tumor size is often the main prognostic indicator in node-negative breast cancer. This variable is particularly important in the decision to use or not to use adjuvant systemic therapy in patients with node-negative breast cancer. Tumor size refers only to the invasive component and should be determined in all three dimensions by the pathologist. Approximately 25 to 30% of patients with negative lymph nodes and a primary tumor less than 2 cm in diameter will experience a recurrence within 20 years of follow-up.^{272, 308, 325, 326}Patients with tumors 1 cm or less in diameter have an excellent prognosis, with fewer than 15% recurring at 10 years.^{326, 327} The largest database demonstrating the relationship among tumor size, lymph node status, and breast cancer survival comes from the Surveillance, Epidemiology, and End Results (SEER) program (see Table 121-9).³²⁸ Less than 2% of patients with tumors under 1 cm and negative nodes died of breast cancer within 5 years. Considering the excellent prognosis for this group of patients with very small tumors, as well as the expense and toxicity of treatment, routine use of chemotherapy is not indicated. The combination of poor nuclear grade and lymphatic vessel invasion identifies a small subset (approximately 10%) of patients with T1a, b N0 M0 breast cancer with a significant risk of relapse, in the 30 to 40% range, that warrants systemic adjuvant therapy.^{5, 329}

Histologic Variables

Several histologic variables have been reported to have prognostic significance (Table 121-10). Most invasive breast cancers are ductal not otherwise specified (NOS).⁵ The prognoses of ductal and lobular carcinomas are similar enough to prompt the same treatment modalities. However, there are several cancers that are less common but have more favorable prognoses.³³⁰ These include pure tubular carcinoma, mucinous or colloid carcinoma, papillary carcinoma, and all noninvasive breast cancers. These cancers have substantially better prognoses, especially when found in a node-negative stage.^{8, 293, 331, 332} The more favorable prognosis of these histologic types often justifies omission of adjuvant systemic treatment, especially for small tumors (< 3 cm). Since most of these special types have small dimensions when diagnosed and are node negative, regional treatment is usually all that is required. When stringently defined, medullary carcinoma, when associated with negative axillary lymph nodes, is also considered by some but not all experts to have a better prognosis.^{6, 332–334} The magnitude of this difference, in terms of 10-year survival, was reported to be 17%.⁶ However, atypical medullary carcinomas (those that do not fulfill the necessary histological criteria) or mixed medullary and ductal carcinomas have prognoses similar to those for the common varieties of ductal and lobular breast cancers.⁶

Table 121-10

Histologic Variables Associated with a Lower Risk for Recurrence.

Histologic Grade or Differentiation

In general, tumors expressing features that indicate a high degree of tumor differentiation have the most favorable prognosis. The literature is inconsistent, however, on the prognostic importance of many of these variables, possibly because of the subjectivity of histologic assessment, the retrospective nature of most studies, and the lack of sophisticated statistical analysis. The clear definition of various histologic differentiation grades led to the recognition that those grades had reproducible prognostic significance.³³⁵ Trained, experienced breast pathologists can recognize poorly differentiated, moderately well-differentiated, and differentiated tumors without much difficulty; the assessment of the histologic grade for those tumors is quite reproducible.³³⁵ Furthermore, although there is some variation among different pathologists, the concordance rate is quite acceptable.³³⁶ Recent reports based on the SEER tumor registry provided strong evidence supporting the prognostic value of histologic grade determined by “average” pathologists.³²⁶ Something similar can be said for nuclear grade, although some find that histologic grade is a more reliable prognostic indicator because it includes cellular and tissue-related criteria. Well-differentiated histologic types, such as tubular, papillary, or colloid (mucinous), have a lower incidence of axillary nodal metastases and a lower risk of distant recurrence than the more common infiltrating ductal carcinomas NOS.^{329, 337} It should be emphasized that pure noninvasive ductal (DCIS) or lobular carcinomas (LCIS) have very low risk for axillary lymph node involvement and virtually no risk for distant metastases. These lesions do not require systemic adjuvant therapy. However, recent results from clinical trials suggest that the addition of tamoxifen to optimal locoregional therapy reduces the risk of local recurrence and the incidence of second primary breast cancer after noninvasive cancers. The histologic and nuclear grades of tumors have been consistently reported to have prognostic significance, especially in patients with negative nodes.^{337, 338} Less than 20% of such patients having well to moderately well-differentiated (grade 1 or 2) tumors experience a recurrence in 5 years, compared to more than 30% for those with poorly differentiated tumors. Similar results have been reported with nuclear grade. Nuclear grade can be determined in cytologic specimens. Smaller tumors are more often well differentiated, whereas larger tumors are predominantly poorly differentiated. Tumor differentiation is also associated with other prognostic indicators such as ER expression, PR expression, ploidy, and S-phase fraction (SPF).

Angiogenesis Markers

In recent years, it has become evident that angiogenesis plays a substantial role in the growth and spread of malignant tumors.³³⁹ Consequently, markers of angiogenic activity have received increasing attention. Counting the number of tumor capillaries immunohistochemically after staining for factor VIII has been shown to have major prognostic value. Tumors with fewer capillaries have a lower metastatic potential and better prognosis.^340–345 Variations in methodology have led to conflicting results, although the weight of the evidence favors the strong prognostic value of angiogenesis. Markers of angiogenesis have also become putative therapeutic targets.³⁴²

Markers of Proliferative Capacity

Measurement of the proliferation rates of malignant tissues has strong prognostic value for several types of cancer, including breast cancer. Several techniques are used to evaluate the proliferative capacity of the malignant cell, including mitotic indices, thymidine-labeling indices (TLIs) and SPF, and the measurement of proteins expressed preferentially during active phases of the cell cycle. The mitotic index is determined by counting mitotic figures using light microscopy on a tumor specimen stained with hematoxylin and eosin. It is a technique well within the capabilities of any surgical pathology laboratory, and it has been validated by both univariate and multivariate analyses.^346-348 Although the TLI has also been proven to be effective and reproducible, it is much more labor intensive and, for this reason, has not gained general acceptance.^{302, 349} The most commonly used method of evaluating proliferation capacity in the United States is the determination of SPF by flow cytometry.^{289, 350} This technique can be performed on fresh or frozen tissues and on archival paraffin-embedded material. It also provides information about DNA ploidy. For both TLI and SPF, a low value indicates a more slowly proliferating tumor and is associated with a slower rate of recurrence regardless of axillary nodal status. A high SPF is strongly correlated with other adverse prognostic factors, such as high histologic and cytologic grades, aneuploidy, and a negative steroid receptor status. In experienced hands, when using cutoff points validated by the laboratory's own experience, the SPF has independent prognostic value (Table 121-11). The cumulative data suggest that SPF is of greater prognostic value than ploidy status. Multivariate analyses indicate that SPF is one of the most important prognostic factors in primary breast cancer. Many experts use SPF determination for treatment-related decisions, especially in the case of axillary lymph node-negative breast cancer. However, there is no general agreement on the clinical usefulness of SPF as a criterion to determine whether adjuvant therapies are warranted.

Table 121-11

Evaluation of the Prognostic Value of S-Phase Fraction (SPF) Performed by DNA Flow Cytometry in Patients with Negative Axillary Lymph Nodes.

Many proteins play a role in the control of the cell cycle or are expressed at higher levels during certain phases of the cell cycle. Ki-67, MIB1, and proliferating cell nuclear antigen (PCNA) are additional markers for the proliferation rate of malignant tumors.^{299, 351–354} Of these, Ki-67 has been more extensively studied, and it correlates strongly with the results of SPF determination and, therefore, long-term prognosis. The usefulness of PCNA requires additional evaluation. Other proteins, including mitosin, Ki-S1, and the various cyclins, are currently under evaluation.^355–357 Cylin E has been shown to be exceptionally accurate as a prognostic indicator. ^357a

Steroid Hormone Receptors

Both the ER and the PR have been studied extensively in patients with primary breast cancer.²⁹² Both clearly have prognostic value, although their ability to discriminate between low- and high-risk patients is quite limited. Patients with ER-positive tumor tend to have a more indolent course and to metastasize preferentially to soft tissue and bone; conversely, those with ER-negative tumors relapse earlier, and metastases to liver, lung, and central nervous system are more likely. ER-positive tumors are more often well differentiated and are associated with other favorable prognostic characteristics. Although patients with ER-positive tumors tend to have better short-term disease-free and overall survival rates than do patients with ER-negative tumors, the differences between the two groups tend to diminish or even disappear with time.³⁵⁸ Although high levels of receptor expression are predictors of a favorable response to endocrine therapy, some studies have suggested that very high levels of receptor may be associated with worse prognosis, compared to lower ER concentrations.³⁵⁹ In patients with negative axillary lymph nodes, ER status is a weak discriminant between high- and low-risk patients.³⁶⁰ The differences in outcome between the ER-negative and ER-positive, node-negative groups are of insufficient magnitude to base treatment decisions on receptor data alone. Receptor data may be of greater value when combined with other prognostic factors.

PR appeared in some studies to be a more valuable prognostic indicator than the ER. Evaluation of both receptors, previously performed mostly by ligand binding assays, is currently done by IHC or immunocytochemical methods with a high degree of reliability. The best use of steroid hormone receptors is not in the determination of prognosis but in the prediction of response to systemic endocrine therapy and, therefore, the selection of optimal adjuvant systemic treatments.

Molecular Genetic Alterations

It is well accepted that malignant tumors develop as a consequence of multiple critical gene abnormalities. These occur because of oncogene activation, loss of function of tumor suppressor genes, or alterations of other genes critical for cell control mechanisms. Two members of the type I growth factor receptor family, the EGFR (or HER1) and HER-2/neu, are frequently amplified and/or overexpressed in breast cancer.^{256, 361} Overexpression of the proteins encoded by these genes is associated with a more aggressive clinical course, including a higher risk of developing metastases and more rapid growth and tumor progression.^{362, 363} HER-2 is overexpressed in 20 to 30% of breast cancers and overexpression is almost always the result of gene amplification. Overexpression is inversely correlated with ER expression and associated with poorly differentiated tumors with high growth rate. Survival is shorter for patients with HER-2-overexpressing tumors than for patients with normal HER-2 expression. HER-2 overexpression may be associated with resistance to some therapeutic agents, although there is considerable controversy about this aspect.^364-367 Several recent reports suggested, however, that patients with HER-2-overexpressing breast cancer derive greater benefit from anthracycline-containing adjuvant chemotherapy than from other regimens, particularily at higher dose.^{252, 254, 367, 368} Whereas some studies suggested that HER-2 overexpression is a marker of tamoxifen resistance, other analyses have provided conflicting results.^369–371 Currently, there is no reason not to use tamoxifen for the management of patients with ER-positive breast cancer, regardless of HER-2 status. The most important clinical application of HER-2 testing is to identify candidates for HER-2-directed therapy with trastuzumab (Herceptin).^{241, 246} Recent reports have suggested that fluorescence in situ hybridization (FISH) is the most definitive testing method to predict response to trastuzumab. The prognostic value of EGFR overexpression is less well established, and there is no Food and Drug Administration (FDA)-approved diagnostic test or therapeutic intervention that targets this receptor.

P53 is frequently mutated in breast cancers.^271–273 While some studies have found that p53 mutation is associated with poor prognosis and others have found correlations with response or resistance to cytotoxic or endocrine agents, no compelling evidence has emerged yet from large, prospective multivariate analysis to support the routine testing for p53 gene mutation or abnormal protein expression as either prognostic or predictive factor. Evidence is even less compelling to support testing for myc, ATM, RB, or others. Similarly, while the presence of mutations in BRCA1 or BRCA2 is clearly associated with markedly increased risk of developing breast cancer, the prognostic value of gene mutations for these two genes remains to be established.

Bone Marrow Involvement

Several investigators have presented data to suggest that the presence of microscopic involvement of the bone marrow, detectable by sensitive IHC analyses, is a major determinant of prognosis.^372–374 In fact, some have suggested that this finding is superior to axillary lymph node involvement in its prognostic capability.³⁷⁴ While these data are increasingly interesting, this marker of prognosis has not been generally accepted or adopted in clinical practice. There is also ongoing work with detection of circulating cancer cells and its correlation with prognosis or response to systemic therapy.

Several other histologic factors, including lymphatic invasion, vascular invasion, tumor necrosis, and mononuclear cell infiltration, have been associated with better or worse prognoses in at least one report. Although extensive clinical experience supports the prognostic value of each of these factors, they have not survived the test of independence in multivariate analysis, nor have they been sufficiently evaluated to date.

Investigational Markers

Many biochemical, molecular, genetic, and immunologic markers have been under investigation for the past several years. All of them have been suggested by one or more reports to have prognostic significance. However, evaluation of these markers falls short of determining whether they are independent prognostic factors or valuable additions to other proven, commonly used prognostic indicators. Among the promising investigational markers are heat shock proteins, EGFR, p53, nm23, cathepsin D, urokinase plasminogen activators (uPA), urokinase plasminogen activator receptors (uPAR), plasminogen activator inhibitors (PAI-1, PAI-2), Bcl-2, BAX, laminin receptors, and apoptotic rate. Several of these may be shown by additional, well-designed studies to be prognostic markers of clinical relevance.

Some markers can be shown to be relevant as predictors of response to chemotherapy or hormone therapy, whereas others may serve to identify tumors using particular targets of biologic intervention (HER-2/neu, EGFR, etc). Many genes commonly associated with cancer participate in the regulation of cell proliferation and apoptosis. Alterations in these genes may be important in determining the chemosensitivity of cancer cells. For instance, cells that express high levels of cyclin D₁ mRNA and protein have increased resistance to phase-specific agents like methotrexate but not to doxorubicin or paclitaxel.^{354, 357} Mutations or deletions of p53 might be associated with resistance to various DNA-damaging drugs.

Over the years, there have been numerous attempts to reach a consensus on what prognostic factors are of recognized clinical utility. However, with the exception of the basic histopathologic factors, steroid hormone receptors, markers of proliferation, and HER-2 expression, the majority of proposed factors fall short of general acceptance as useful clinical tools.

Prognostic Indices

The large number of prognostic factors available and the incomplete evaluation of most of those factors lead to serious difficulties in the interpretation of prognostic information, especially when caring for an individual patient. There is no standardized method of integrating prognostic information, and attempts to develop prognostic indices have had limited efficacy. The index developed by the breast group at The University of Nottingham is based on tumor grade, axillary lymph node involvement, and hormone receptor status.^{358, 359} It has been validated prospectively and confirmed by an independent center, but it still has limited usefulness in the determination of an individual's prognosis.³⁶⁰ Multivariate analysis is helpful in determining which of several potential factors provides independent prognostic information.^{361, 362} However, the usefulness of this analysis varies with the factors included in (and excluded from) the process, and to date no large-scale analysis of all or even a majority of potential prognostic factors has been performed.

An emerging area of investigation of prognosis and prediction of sensitivity or resistance to therapy is genomics. CDNA arrays, including 20,000 to 30,000 genes are being used to develop gene expression profiles and correlate them with treatment outcome.³⁶³ While this is an evolving technology, it is conceptually attractive, and early results suggest great promise in understanding better the biology of cancers, as well as, determining prognosis and predicting drug sensitivity.

Clinical Use of Prognostic Factors

The use of prognostic factors has two major objectives. One is to calculate the individual risk of recurrence and disease-related mortality for patients with primary breast cancer treated with curative intent. In this group, determination of prognosis will assist in decisions regarding the incorporation of systemic adjuvant therapies into the management of the disease. It used to be considered that patients with a risk of relapse lower than 10% would have such marginal benefit from adjuvant systemic therapies that no such intervention would be recommended. However, recent calculations of the benefits of tamoxifen administered for 5 years to patients with ER-positive breast cancer and the demonstration that the combination of hormonal therapy and chemotherapy has at least additive effects on reduction of risk of relapse and mortality have prompted a reassessment of these recommendations. Most patients with invasive primary breast cancer larger than 1 cm should be encouraged to receive adjuvant systemic therapy; some with tumors smaller than 1 cm should also be advised to consider adjuvant therapy, since randomized trials have not identified any subgroups that failed to derive benefit from these interventions. The use of prognostic factors, under these circumstances, serves more to place the benefit:risk ratio for individual patients in the context of their own prognosis. The identification of subgroups with a very high risk of relapse is useful to determine eligibility for novel or more aggressive interventions. However, the acceptable benefit:risk ratio varies substantially from person to person; therefore, after a best attempt to calculate individual risk of relapse and mortality, treatment options and the probability of benefit from each should be presented in the light of toxicity and cost to facilitate informed treatment decisions.