Uncommon breast tumors in perspective: Incidence, treatment and survival in the Netherlands



The relatively small group of patients with breast tumors other than the ductal, lobular or mixed ducto-lobular types, has reached nonnegligible numbers due to the ongoing increase in the incidence of breast cancer. We investigated stage and grade distribution of uncommon breast tumors using the nation-wide Netherlands Cancer Registry (population 16.5 million) and incidence patterns, treatment and long-term survival (up to 19 years) using the regional Eindhoven Cancer Registry (population 2.4 million). Incidence of all uncommon breast tumors together was 9.2/100,000 person years (age-standardized, ESR). The proportion of stage I tumors was 70% among patients with tubular (n = 3,456) and 40–50% for mucinous (n = 3,482), papillary (n = 1,078), cribriform (n = 503) and neuroendocrine (n = 76) tumors, contrasting to 27, 28 and 36%, respectively among patients with Signet ring cell cancer (n = 75), Paget's disease (n = 818) and the common invasive ductal carcinomas (n = 121,656). A better age-, stage-, and grade-adjusted prognosis was observed for patients with lobular (death risk ratio 0.8, 95%CI: 0.7–0.9), mucinous (0.5, 0.3–0.9), medullary (0.5, 0.3–0.9) and tubular (0.4, 0.2–0.6) carcinoma or phyllodes tumor (0.02, 0.0–0.2), compared with invasive ductal carcinomas. For patients with papillary (0.6, 0.2–1.6) and cribriform (0.1, 0.0–5.1) tumors better prognosis was not statistically significant. In conclusion, histologic type was an essential determinant of survival for about 10% of all newly diagnosed women with invasive breast cancer. Because patients with mucinous, tubular, medullary and phyllodes tumors have such a good prognosis, less aggressive treatment should be considered in some cases whereby specific guidelines are becoming increasingly desirable. Communication to patients with these specific histological types should reflect this. © 2007 Wiley-Liss, Inc.

The large majority of breast cancer patients is diagnosed with an invasive ductal (about 65–75%) or lobular (15%) carcinoma.1 An increasing proportion of up to 7% is diagnosed with a mixed ducto-lobular tumor (2-fold increase between 1987–1999 in the USA).2 The remaining 10% of newly diagnosed breast cancers with another histology have been the subject of only a limited number of studies.

Previously, the substantial variation found in age-specific incidence between different histological subtypes in large population-based studies,3, 4 suggested different aetiologies. Patients with a papillary or mucinous tumor are 5–10 years older than those with a ductal carcinoma, while patients with a medullary tumor tend to be about 10 years younger.3, 5 Survival according to morphology has been reported in a few studies, but these described small series of patients,6–9 only 1 or 2 specific subtypes,6, 9, 10 on patients diagnosed and treated before 1995,6, 8, 11 or without complete information on treatment.12, 13 In these studies prognosis for mucinous,10, 12, 13 tubular,7, 10, 13 medullary13 or papillary13 carcinoma was significantly better than that of patients with an invasive ductal carcinoma.

Within the increasing incidence of breast cancer the incidence and the number of patients with an uncommon tumor may increase more. Therefore, more knowledge about the frequency and natural history of these tumors is desirable. A large population is necessary to obtain a sufficient number of patients with these relatively rare tumors in a reasonable time span.

The Netherlands Cancer Registry (NCR), in operation since 1989 and covering the whole country with a current population of 16.5 million people, provides a unique opportunity to investigate the occurrence and characteristics of these uncommon breast tumors. In addition, we will study details regarding treatment and long-term survival in a part of the NCR, the Eindhoven Cancer Registry (ECR) that was started in 1955.

Patient and methods

We used data from the nation-wide NCR (www.ikcnet.nl), which consists of 9 regional cancer registries since 1989. The cancer registries receive lists of newly diagnosed cases on a regular basis from the pathology departments, all participating in a nation-wide pathology network (PALGA). In addition, the medical records in departments of hospitals provide lists of diagnoses of outpatients and hospitalized cancer patients. Following this notification, trained tumor registration clerks abstract information according to a minimum data set, including patient characteristics (gender, birth date), tumor information (topography, histology, stage, date of diagnosis) and treatment.

Topography and histology were coded according to the International Classification of Diseases for Oncology (ICD-O).14 We grouped the histology codes according to the classification in Table I.

Table I. Classification of Morphology of Consecutive Patients Diagnosed with Malignant Breast Tumors in the Netherlands (1989–2003) and Southeastern Netherlands (1973–2004)
  Netherlands Cancer RegistryEindhoven Cancer Registry
Histological groupMorphology code according to ICD-O14n%n%
  • 1

    Only patients with invasive lymphoma of the breast stage IE (Ann Arbor).

Invasive ductal (not otherwise specified)8010–8021,8140–8141,8190,8230–8231,8310–8441, 8500–8502,8508,8514,8521,8523,8530,8550121,6567722,16078
Mixed ducto-lobular85226,2313.99613.4
Metaplastic8031–8035, 8560,8570–8573, 8575, 8801,8980, 9180,9183,92203120.2330.1
Squamous cell (SCC)8070–8074440.0120.0
Adenoid cystic8200980.190.0
Neuroendocrine (carcinoid)8240–8246,8574,8249,9091760.050.0
Signet ring cell8490750.050.0
Paget disease8540–85438180.51270.4
Sarcoma8800,8810–8921,8931,8935,8990,8991,9040–9044, 9131–9179, 9184–9219, 9221–9230, 9232–9342,9364–9372, 9473,9540–9559, 9561–9581790.0150.1
Otherall other codes, including patients without pathological verification1,3730.93441.2
Subtotal uncommon13,6668.62,1337.5
Total 158,84610028,359100

We included all patients diagnosed with a first histological type of malignant breast tumor between January 1989 and December 2003 (n = 158,353). In addition, 493 patients with a second primary breast tumor other than invasive ductal, lobular, mixed ducto-lobular or other/NOS were included in the analyses, resulting in 158,846 tumors for analyses of which 13,666 (8.6%) uncommon histological types.

These data were used to analyze the age at diagnosis and the stage and grade distribution according to histological group.

Age was presented as median, with 25 and 75 percentiles (Q1–Q3) and the 1–99% range.

Stage of the solid tumors was categorized according to the IUCC TNM-classification15 for all patients diagnosed until 2002. From 2003 onwards, when a new classification became effective,16 patients with more than 3 positive lymph nodes, previously included in stage II, are now considered stage III. Stage could not be classified according to the TNM system for patients with (hemangio) sarcomas, phyllodes tumors and those with other/NOS tumors (the latter groups comprised 403 patients (0.25%) with a clinical diagnosis only). For patients with a lymphoma the Ann Arbor classification was used. We selected only patients with Ann Arbor stage IE, i.e. the patients with only 1 extranodal tumor site, i.e. the breast17 (n = 98, 0.6% of all lymphomas in women, 4% of all extranodal lymphomas).

Grade was categorized in 4 classes: well (i), moderately (ii) or poorly (iii) differentiated or undifferentiated/anaplastic (iv). Strict criteria for assigning stage were only followed since the most recent decade, so the value of the tumor grade before that time is somewhat limited.

In addition we used data from the ECR in the south of the Netherlands (population 0.9 million between 1975 and 1985, over 2 million since 1988 after an expansion of the registration area, comprising 15% of the Dutch population in 2004). This population-based cancer registry was started in 1955 and is the oldest cancer registry in the Netherlands and among the oldest in Europe. Data on patients who lived in the catchment area of the ECR, but who were diagnosed in hospitals elsewhere in the Netherlands, were retrieved on a routine basis from all other Dutch Cancer Registries since 1989. This has been completed up to 2002, so incidence rates for 2003 and 2004 may be underestimated by about 3% (unselected patients).

From the ECR, all breast cancer patients diagnosed between 1973 and 2004 (n = 28,359) were used to study time trends in incidence. Age-adjustment was performed by direct standardization according to the European Standard Population (European Standardized Rates, ESR). Annual incidence rates were calculated as 3-year moving averages.

Trends in incidence were estimated by calculating the estimated annual percentage change (EAPC). This was done by fitting a regression line to the natural logarithm of the rates using calendar year as regressor variable, i.e., y = mx + b where y = ln(rate) and x = calendar year. Then EAPC = 100 × (em – 1). This calculation assumes that the rates increased or decreased constantly over the entire period.

We used information on type of treatment (surgery, radiotherapy, systemic therapy) for patients diagnosed in 1985–2004 (n = 24,217); these results are presented separately for patients younger and older than 70 years at diagnosis separately.

For a subgroup of patients (n = 16,797) diagnosed in the eastern part of the ECR region and those diagnosed since 1994 in the western part (all until 2002), we obtained complete information on vital status up to January 1, 2005 from the Central Bureau for Genealogy. This registry provides virtually complete coverage of all deceased Dutch citizens.

Relative survival (the ratio of the observed to the expected survival rates) is an estimation of disease-specific survival, which reflects survival of cancer patients adjusted for survival in a reference population with the same age structure.18 Expected survival rates were calculated from life tables for the female population of the ECR catchment area with the same 5-year age distribution. We used generalized linear models with a Poisson error structure, based on collapsed data and exact survival times.19 When the number of patients effectively at risk in a particular category became too small for a reliable estimation of relative survival (i.e. <10), the results are not shown.

All statistical analyses were performed with SAS (SAS System version 9.1, Cary, CA).


The incidence of all uncommon breast cancers combined (i.e. all breast cancers other than invasive ductal, lobular or mixed ducto-lobular) was 9.2/100,000 person years (ESR) over the period 1989–2003, in total more than 13,500 new patients in the Netherlands (8.6% of all breast cancers).

The median age varied according to histological type (Fig. 1). The median age of patients with a ductal carcinoma at diagnosis was 60 years (Q1–Q3: 49–72). Patients with a phyllodes tumor or medullary or tubular carcinoma were younger (median age 52, 53 and 57 years, respectively). Those with other histological types were older; patients with a histology classified as other/NOS were the oldest at diagnosis (median 82 (73–87)).

Figure 1.

Age distribution of consecutive patients diagonised with invasive breast cancer in the Netherlands 1989–2003 (median, Q1, Q3, P1, P99).

Stage distribution was most favorable for patients with mucinous, tubular, papillary, cribriform and neuro-endocrine tumors (Table II). For example, the proportion of patients with a stage I tumor was 42, 70, 45, 52 and 42%, respectively, in comparison with 36% of patients with an invasive ductal carcinoma. A less favorable stage distribution with only 27 and 28% stage I tumors was observed for patients with Signet ring cell cancer and Paget's disease, respectively. Although tumor grade was unknown for a relatively large proportion of patients (44%) the proportion with a grade 1 tumor was high among patients with mucinous, tubular, papillary, cribriform and phyllodes tumors, compared with those with an invasive ductal carcinoma.

Table II. Characteristics of Patients with Malignant Breast Tumors Diagnosed 1989–2003 in the Netherlands According to Histological Group
 Invasive ductal, nosLobularMixed ducto-lobularMucinousMedullaryTublarPapillaryMetaplasticSquamous cellAdenoid cystic
  1. NA = not applicable.

Age at diagnosis                    
 <50 years34,074284,103242,0613345813728431,001291591591297162020
 70+ years36,307305,625331,427232,053592731667820563521093521483536
Stage at diagnosis                    
Total121,656 17,293 6,231 3,482 1,677 3,456 1,078 312 44 98 
 CribriformNeuro endocrineSignet ring cellPaget diseaseSarcomaHemangio sarcomaPhyllodesLymphomaOther/NOSTotal
Age at diagnosis                    
 <50 years12725811121620525243091319345141457443,35127
 70+ years18336425539523003727343651801947481,1098148,95431
Stage at diagnosis                    
Total503 76 75 818 79 70 427 98 1,373 158,846100

The incidence of invasive ductal carcinoma varied between 75 and 85/100,000 person years (ESR) during the 70s and 80s, but has increased markedly since the early 1990s (Fig. 2). In the mid 1990s during the start of the mass screening program for women of 50–69 years (since 1998 extended to women of 70–74 years) a small decline was observed. But thereafter the incidence increased markedly again, to about 110/100,000. The incidence of lobular breast tumors increased markedly until 1985 up to about 12/100,000, thereafter it increased more modestly to about 15/100,000 in 2003, the mixed ducto-lobular tumors increased from 1985 onwards and has been to about 5/100,000 since 1995. In 2004 these 2 groups together comprised about 15% of all invasive breast cancers. The incidence of all uncommon tumors combined was generally low (3 to 5/100,000/year) during the 1970s and 1980s, but increased over time to 12/100,000 in 1998; it subsequently decreased to about 7/100,000 in 2004. A markedly increased incidence in the mid 1990s followed by a decline was observed for patients with tubular, phyllodes and cribriform tumors. This trend was also observed for (hemangio) sarcoma, but about 5 years earlier. The incidence of patients with a medullary tumor has decreased since the early 1990s from 1.8 to 0.6/100,000 in 2004 (EAPC: −9.4%, p = 0.001). A steady increase, also in recent years, was observed for papillary (EAPC: +7.5%, p = 0.001) and metaplastic (EAPC: +10%, p = 0.05) tumors and lymphomas (EAPC: +6%, p = 0.2).

Figure 2.

Trend in incidence (3-year moving averages) according to histology for consecutive patients diagonised with invasive breast cancer in southeastern Netherlands 1973–2004.

The large majority of patients underwent surgery, especially those younger than 70 (94%) (Table III). Treatment of patients below age 70 with mucinous, tubular, papillary, cribriform, Paget or (hemangio) sarcomas was generally less aggressive; the proportion receiving adjuvant systemic therapy in addition to surgery and radiotherapy was significantly lower (respectively 15, 13, 24, 22 and 0%) in comparison with patients with invasive ductal carcinomas (30%). Similar tendencies were observed for patients older than 70 years at diagnosis. Systemic therapy (mostly hormonal therapy) was administered to 48% of patients with an invasive ductal carcinoma, these proportions were 34, 34, 28 and 19%, respectively, for patients with mucinous, tubular, papillary and cribriform carcinoma.

Table III. Treatment of Patients with Malignant Breast Cancer Diagnosed in the Southeast of The Netherlands 1985–2004 According to Histological Group and Age at Diagnosis
Age <70Invasive ductal, nosLobularMixed ducto-lobularMucinousMedullaryTubularPapillaryMetaplasticSquamous cellAdenoid cystic
  1. S = surgery, RT = radiotherapy, ST = systemic therapy.

S alone2,4181837419154213721392390211831213233117
S + RT4,92436703352122992516638239572645425117583
S + RT + ST4,07530586292463326154828561361074411700
S + ST1,6091224212951315815923561021311700
ST alone24223924100111000000000
Total13,792 2,005 742 179 172 422 58 16 6 6 
 CribriformNeuro endocrineSignet ring cellPaget diseaseSarcomaHemangio sarcomaPhyllodesLymphomaOther/NOSTotal
S alone819001333142444457599217123,24418
S + RT1433110001622556004600366,31536
S + RT + ST102400133162200000017005,07929
S + ST717001339120022900320002,03011
ST alon000000000000001722433102
Total42 1 3 73 9 7 64 15 51 17,663100
Age 70+Invasive ductal, nosLobularMixed ducto-lobularMucinousMedullaryTubularPapillaryMetaplasticSquamous cellAdenoid cystic
S alone1,124241962362287634122924251429747375267
S + RT1,06022199235224582614333840204242700133
S + RT + ST907191752053241995121718613000000
S + ST86118156184320431961414156132130000
ST alon544119811211362511120012500
Total4,766 859 218 221 42 96 48 15 4 3 
 CribriformNeuroendo crineSignet ring cellPaget diseaseSarcomaHemangio sarcomaPhyllodesLymphomaNOSTotal
S alone1257375002146360510686225111,57324
S + RT314125007151200000113001,45922
S + RT + ST3140015081700000000001,19418
S + ST15000081700000000211,14217
ST alon00001501212000114001387580412
Total21 4 2 46 5 5 7 8 184 6,554100

Relative 5-year survival rates for patients with lobular (82%) or mixed ducto-lobular (84%) carcinomas were slightly more favorable than those for patients with an invasive ductal carcinoma (81%) (Fig. 3). Prognosis was significantly worse for those with Paget's disease, metaplastic tumors, breast lymphoma or other/NOS tumors, with 3-year survival rates of 75, 59, 69 and 54%, respectively. All other histological groups exhibited significantly better survival rates compared with invasive ductal carcinoma, especially patients with tubular carcinoma or phyllodes tumor, with 5-year relative survival rates of 97 and 94%, almost similar to that of women without breast cancer. Survival of patients with cribriform tumors was comparable with that of women never diagnosed with breast cancer (5-year relative survival 104%).

Figure 3.

Relative survival according to histology for consecutive patients diagnosed with invasive breast cancer in the southeastern Netherlands 1985–2004.

Multivariate analyses of relative survival (Table IV) showed a significantly better prognosis for patients with lobular (relative excess risk of death (RER) 0.82, 95%CI: 0.7–0.9), mucinous (0.53, 0.3–0.9), medullary (0.52, 0.3–0.9), tubular (0.39, 0.2–0.6) carcinoma or phyllodes tumor (0.02, 0.0–0.2), independent of age at diagnosis, stage and grade. Prognosis for patients with papillary (RER 0.57, 95%CI: 0.2–1.6) and cribriform (0.11, 0.0–5.1) tumors was also better than that for patients with invasive ductal carcinoma, though not statistically significant.

Table IV. Multivariate Regression Analysis of Relative Survival Rates of Consecutive Patients with Malignant Breast Tumors in Southeastern Netherlands, 1985–2002
 RER95% CI
  • NA = not applicable.

  • 1

    Number of patients at risk too small for calculation of RER.

Histolgical group  
 Invasive ductal, nos1.00 
 Mixed ducto-lobular0.950.8–1.1
 Squamous cell carcinoma (SCC)1  
 Adenoid cystic1  
 Signet ring cell1  
 Paget disease0.760.4–1.3
Age at diagnosis  
 <50 years1.00 
 50–69 years1.041.0–1.1
 70+ years1.361.2–1.5
Stage at diagnosis  


The number of patients with uncommon tumors in the Netherlands has increased markedly since 1989, with about 700–800 new patients every year; for this largely heterogeneous group the incidence lies between those of cervical and ovarian cancer.

The less frequent histological types of breast cancer were clearly other entities than ductal carcinoma, as shown by differences in age at diagnosis, stage and grade distribution. The pattern of incidence rates over time was different from that of ductal carcinoma and relative survival rates also differed for the uncommon tumors, with very good survival rates for patients with cribriform or tubular cancer or phyllodes tumor, and poor survival for the lymphomas and metaplastic tumors.

Use of data from the large nation-wide NCR allowed analyses of rare tumors which otherwise could not have been studied at a population-based level due to insufficient numbers. A potential drawback might be the involvement of a large number of pathologists who diagnosed the breast tumors, which may have led to some morphology misclassification, although it seems unlikely that this resulted in over- or under representation of specific histological types. This is also supported by the similarity of histological distribution of the Dutch (NCR) and the ECR. Also, the distribution according to age, stage and grade between the 2 populations was very similar (data not shown). Furthermore, the pathology diagnoses became more accurate over time, the number of pathologists increased, especially since the 1980s and they became increasingly aware of differences in histological types. This increased awareness is probably related to a decrease in the incidence of breast cancers with unknown histology. Simultaneously with this decrease, the incidence of all specific histological groups (including invasive ductal NOS, lobular and mixed ducto-lobular) increased and the total incidence of unknown histologies were lower than the incidence of all uncommon tumors combined. Thus, it seems unlikely that the majority of uncommon tumors were previously diagnosed with unknown histology.

The median ages at diagnosis for the different histological types were similar to the population-based SEER data on patients diagnosed in the 1990s4 and also to other histological type-specific studies in industrialized countries.7, 9, 20, 21 Patients with medullary breast cancer are generally younger, possibly because this type of tumor might be hereditary, occuring at a younger age than sporadic breast tumors.22

The stage and grade distribution showed the less aggressive nature of mucinous, tubular, papillary and cribriform carcinoma and phyllodes tumors, as also found in earlier studies.4, 7, 9, 10, 12

Women diagnosed with tubular carcinoma are more likely to be detected at screening.23 Our results also suggest this, because the incidence of tubular carcinoma increased markedly in the mid 1990s followed by a decline a few years later. This is in accordance with the introduction of population screening for women aged 50–69 in the area of the ECR in 1991 which was fully implemented in 1996. This temporary rise in incidence was also observed for phyllodes tumors and cribriform carcinoma. A few years earlier, an increase followed by a decrease was observed for (hemangio) sarcomas, which are related to previous radiotherapy to the breast.24, 25 In our study 28 patients (40%) with an hemangiosarcoma had suffered a prior breast cancer. The increasing incidence rates for lobular and mixed ducto-lobular carcinoma might be related to the use of hormone replacement therapy,2, 26–28 although this use remained relatively modest in the Netherlands (10–16%29–31 compared with over 50% in the USA32).

Treatment of uncommon breast tumors has rarely been studied, because it is difficult to obtain sufficient numbers of patients in each category. In an average hospital one observes most of these categories of patients only once every 5–10 years. So, it is not illogical that in the Dutch guidelines for breast cancer treatment, no specifications according to histological type are shown (although it is stated that axillary treatment of patient with tubular carcinomas smaller than 1 cm is not necessary),33 nor were they an issue at the international expert consensus meeting in St. Gallen (Switzerland).34 Therefore treatment decisions were probably based on the same criteria as for invasive ductal carcinoma: stage, grade, menopausal status and hormone receptor status. Unfortunately, we do not have information about the latter in our database. We observed less aggressive treatment of patients with mucinous, tubular, papillary and cribriform tumors in accordance with the favorable stage and grade distribution. However, for older patients comorbidity may also have played a role.35 Treatment recommendations for patients with some rare tumors, e.g. adenoid cystic,36 tubular37 and phyllodes,38 advise excision, in some cases combined with radiotherapy but without adjuvant chemotherapy. Patients with metaplastic tumors should be treated as invasive ductal carcinomas,39 and those with lymphoma of the breast should be treated as all extranodal lymphoma patients, based on appropriate staging and histopathology.

Like previous studies (Table V), we found a very good prognosis for patients with tubular,7, 10, 13, 40 and mucinous9, 10, 12, 13, 40 carcinoma. Relative 5-year survival of patients with medullary cancer was equal to about that found in a study which combined data over 1990–1992 from 6 European cancer registries.8 Because of the limited number of cases, only a few studies have reported on the survival of patients with cribriform carcinoma.41, 42 In our relatively recent series of patients the prognosis was even better than for women in general. The very good prognosis for patients with papillary carcinoma has been reported before from the SEER database.13, 40

Table V. Overview of Studies of Patients with Uncommon Tumors of the Breast
AuthorsStudy populationHistological typesNo. of patientsSurvival rates1Risk ratio2Median survivalRemarks
  • 1

    crude survival rate unless otherwise stated.

  • 2

    compared to invasive ductal carcinoma, nos.

Sullivan et al.7Massachusets General Hospital 1980–2002tubular73  90.5 months 
Allemani et al.8Population-based cancer registries of Estonia, France, Italy, Spain, Netherlands, UK 1990–1992medullary special types (i.e. tubular, apocrine, cribriform, papillary, mucinous, signet ring cell)625-yr: 86% (76–96)1.19 (0.49–2.85) relative survival, RER, adjusted for age, stage, hormone receptor status
2115-yr: 95% (90–100)0.35 (0.12–0.99)
Ellis et al.40Nottingham City Hospital 1974–1987tubular3810-yr: 90%   
cribriform1310-yr: 91%
mucinous1410-yr: 80%
medullary4410-yr: 51%
Komenaka et al.9Columbia University-Presbyterian Medical Center 1980–1998mucinous655-yr: 93.6%   
10-yr: 72.8%
Kuper-Hommel et al.412 regional population-based cancer registries in the Netherlands 1981–1999lymphomas382-yr: 63% 38 months 
Li et al.13SEER 1974–1998mucinous3,923 0.80 (0.86–0.91) Only patients 50–79 yrs HR, adjusted for age, year of diagnosis, stage, SEER registry, surgery, RT
medullary2,9020.82 (0.78–0.87)
tubular2,2600.66 (0.60–0.73)
papillary1,0490.81 (0.73–0.90)
Fu et al.20Providence Hospital and Medical Centers 1980–1999Paget41  42 months 
Diab et al.10San Antonio, TX, breast cancer databasestubular4445-yr: 88% (p < 0.001)  p for comparison with ductal carcinomas nos
mucinous1,2215-yr: 80% (p = 0.088)
Ha et al.42University of Texas M.D. Anderson Cancer Center Lymphoma database 1972–1994lymphoma235-yr: 74%   
Northridge et al.12SEER 1973–1990mucinous4,082 0.38 (0.34–0.42) HR, adjusted for age, stage, year of diagnosis, race, grade
Gamel et al.11SEER 1973–1991medullary2,908  2.8–3.9 yr 
Berg and Hutter40SEER 1973–1987medullary4,4865-yr: 82%  relative survival
mucinous3,5535-yr: 95%
paget1,7755-yr: 79%
papillary1,3955-yr: 95%
tubular1,0925-yr: 96%
Pederson et al.6Danish Breast Cancer Cooperative group 1977–1987medullary2355-yr: 82%  read from the graphs
10-yr: 77%
Venable et al.44George Washington University Medical Center 1971–1975, 1981–1986cribriform325-yr: 100%  12 patients with pure cribriform, 20 with at least 50% cribriform component
Wargotz and coworkers45, 46Armed Forces Institute of Pathology Washington, DC before 1983metaplastic: matrix-producing265-yr: 68%   
spindle cell1005-yr: 64%
Lattes471970slymphoma335-yr: 9%   

The poor survival rates for patients with metaplastic tumors is probably affected by the heterogeneity of the group; previous studies consisted of small series of cases and 5-year survival rates ranged from 401 to 68% for the matrix-producing subtype.43 Survival of patients with breast lymphomas has been reported to be very poor,44, 45 but survival might improve for more recent patients through modern staging and therapy.46

RERs of death were calculated for most histological types, if a sufficient number of patients were available. Only one previous study reported RERs according to histology8 but could not distinguish the various histologic subtypes, nor could it adjust for grade. We found that histology significantly predicted survival after adjustment for age, stage and grade. Treatment, which varied over the years and is largely related to stage and grade, was also added to the model, but this did not change the RERs significantly (data not shown).

In conclusion, in about 10% of all newly diagnosed breast cancer patients specific histology has implications for the detection, diagnosis, stage (and thus also treatment) and survival. Despite a less aggressive treatment, the survival for certain histological types appeared to be very high, even comparable with that of women without breast cancer. Furthermore, after adjustment for age, stage and grade, patients with mucinous, tubular and medullary carcinoma or phyllodes tumors exhibited such a low risk of death that in some cases even less aggressive treatment should be considered. Communication to patients with these specific histological types should reflect this relatively favorable prognosis.


We thank Prof. J.W.R. Nortier, Department of Medical Oncology, Leiden University Medical Center, Leiden and Dr. E. Rutgers, Department of Surgery, The Netherlands Cancer Institute, Amsterdam, for their valuable comments on the manuscript.