Journal of Neuro-Oncology

, Volume 112, Issue 3, pp 467–472 | Cite as

The effect of tumor subtype on the time from primary diagnosis to development of brain metastases and survival in patients with breast cancer

  • Paul W. Sperduto
  • Norbert Kased
  • David Roberge
  • Samuel T. Chao
  • Ryan Shanley
  • Xianghua Luo
  • Penny K. Sneed
  • John Suh
  • Robert J. Weil
  • Ashley W. Jensen
  • Paul D. Brown
  • Helen A. Shih
  • John Kirkpatrick
  • Laurie E. Gaspar
  • John B. Fiveash
  • Veronica Chiang
  • Jonathan P.S. Knisely
  • Christina Maria Sperduto
  • Nancy Lin
  • Minesh Mehta
Clinical Study

Abstract

Our group has previously published the Diagnosis-Specific Graded Prognostic Assessment (GPA) showing the prognostic factors associated with survival in patients with brain metastases (BM). The purpose of this study is to investigate the relationship of breast cancer subtype to the time interval from primary diagnosis (PD) to development of BM (TPDBM), number of BM at initial BM presentation and survival. We analyzed our previously described multi-institutional retrospective database of 865 breast cancer patients treated for newly-diagnosed BM from 1993 to 2010. Several factors found to be associated with survival were incorporated into the Breast-GPA, including tumor subtype. The GPA database was further analyzed to determine if the subtype correlated with the TPDBM, number of BM, and survival from PD. After exclusions for incomplete data, 383 patients remained eligible for analysis. The subtypes were approximated as follows: Luminal B: triple positive; HER2: HER2 positive/ER/PR negative; Luminal A; ER/PR positive/HER2 negative; Basal: triple negative. Patients with Basal (90), HER2 (119), Luminal B (98) and Luminal A (76) tumor subtypes had a median TPDBM of 27.5, 35.8, 47.4 and 54.4 months (p < 0.01), median survival from PD of 39.6, 66.4, 90.3 and 72.7 months (p < 0.01) and median survival from BM of 7.3, 17.9, 22.9 and 10.0 months (p < 0.01), respectively. Tumor subtype is an important prognostic factor for survival in patients with breast cancer and BM. Although TPDBM is not an independent prognostic factor for survival (and thus not part of the Breast-GPA), the TPDBM does correlate with tumor subtype but does not correlate with the number of BM. Patients with Basal and HER2 tumor subtypes have short TPDBM. Prospective studies are needed to determine if screening brain MRIs are indicated in patients with Basal or HER2 subtypes.

Keywords

Breast cancer Brain metastases Prognosis Radiation therapy Stereotactic radiosurgery Estrogen Progesterone HER2 Graded prognostic assessment 

Introduction

Breast cancer is the most common cancer and the second leading cause of cancer-related mortality in women in the USA. An estimated 229,600 women were diagnosed with breast cancer and an estimated 39,920 died from the disease in 2012. [1] Brain metastases occur frequently in women with breast cancer. An estimated 10–30 percent of women with breast cancer will develop brain metastases. [2] While the overall prognosis of patients with brain metastases remains poor, we now understand that this population is markedly heterogeneous. In 1997, the Radiation Therapy Oncology Group (RTOG) Recursive Partitioning Analysis (RPA) established a prognostic index for patients with brain metastases which was helpful in guiding patient care and designing clinical trials. [3] That index has been updated to the Graded Prognostic Assessment (GPA) (Table 1). [4] Prognostic factors found to be significant by either multivariable Cox regression (MCR) or recursive partitioning analysis were retained in the final MCR model to improve its prognostic ability. The relative magnitudes of the regression coefficients (i.e. log hazard ratios) from the final model were used to design and weight the GPA, an additive point-based prognostic index. A score of 4.0 correlates with the best prognosis, and a score of 0.0 correlates with the worst prognosis. The data supporting the derivation of the GPA demonstrate that the prognosis for patients with brain metastases varies by diagnosis as do the prognostic factors predicting survival. This is especially true in breast cancer where median survival by GPA class varies from 3 to 25 months after the development of brain metastases, depending on several variables, including tumor subtype. [5] Patients with HER2-positive breast cancer have an increased risk of developing brain metastases [6], and the prognostic importance of tumor subtype is now well established. [7, 8, 9, 10] Improved management of brain metastases has resulted in a shift in the cause of death. Only 15–30 % of these patients dying from neurological causes [11, 12], although in HER2-positive patients treated with trastuzumab, 40 % of patients with BM die of neurological causes; this association of death with neurologic causation is usually weak, because of data quality issues. [13] Systemic disease is the most common cause of death, even in patients with BM.
Table 1

GPA worksheet to estimate survival from brain metastases in breast cancer

Prognostic factor

0

0.5

1.0

1.5

2.0

Score

KPS

≤50

60

70–80

90–100

n/a

Subtype

basal

n/a

LumA

HER2

LumB

Age

≥60 

<60

n/a

n/a

n/a

     

Sum Total = 

Subtype

 Basal = Triple Negative (ER/PR/HER2-neg)

 LumA = Luminal A (ER/PR-pos, HER2-neg)

 LumB = Luminal B (Triple Positive, ER/PR/HER2-pos)

 HER2 = HER2-pos, ER/PR-neg

MST (mo) by GPA

 0–1.0 = 3.4

 1.5–2.0 = 7.7

 2.5–3.0 = 15.1

 3.5–4.0 = 25.3

GPA graded prognostic assessment, KPS Karnofsky performance score, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2

Our group has previously published studies based on a retrospective database of 4,259 patients with brain metastases from 11 institutions. [5, 14, 15] Within that database, there were 400 women with brain metastases from breast cancer with adequate data to define tumor subtype and its association with survival. The purpose of this study is to investigate the relationship of tumor subtype to the time from primary diagnosis (PD) to development of brain metastases (TPDBM), number of brain metastases, and survival from primary diagnosis as well as from treatment of brain metastases.

Methods

Patient population

An institutional review board–approved retrospective database of 865 patients with breast cancer treated for brain metastases between June 1993 and January 2010 was generated from the radiation oncology departments at 11 institutions. Of these 865 patients, 482 were excluded for one or more of the following reasons: recurrent (not newly diagnosed) brain metastases (n = 157), unknown or surgery-only treatment (n = 7), or missing data on survival (n = 29)/tumor subtype (n = 295)/KPS (n = 9). Table 2 shows the patient characteristics for all 400 patients, reported in an earlier study [5], with data adequate to define subtype. An additional 17 patients were excluded because of incomplete data for TPDBM so this analysis is based on the 383 patients.
Table 2

Patient characteristics, overall and by tumor subtype

Factor

Level

N (%)

Tumor Subtype

p value

Luminal Bn (%)

Luminal A n (%)

HER2 n (%)

Basal n (%)

Age (Median = 53)

<50

144 (38 %)

43 (44 %)

20 (26 %)

42 (35 %)

39 (43 %)

0.33

50–59

139 (37 %)

35 (36 %)

31 (41 %)

44 (37 %)

29 (32 %)

60–69

70 (18 %)

14 (14 %)

15 (20 %)

25 (21 %)

16 (18 %)

≥70

30 (8 %)

6 (6 %)

10 (13 %)

8 (7 %)

6 (7 %)

KPS (Median = 80)

<60

16 (4 %)

3 (3 %)

3 (4 %)

6 (5 %)

4 (4 %)

0.47

60

21 (5 %)

5 (5 %)

7 (9 %)

4 (3 %)

5 (6 %)

70

65 (17 %)

21 (21 %)

13 (17 %)

17 (14 %)

14 (16 %)

80

98 (26 %)

20 (20 %)

27 (36 %)

28 (24 %)

23 (26 %)

90

141 (37 %)

38 (39 %)

19 (25 %)

52 (44 %)

32 (36 %)

100

42 (11 %)

11 (11 %)

7 (9 %)

12 (10 %)

12 (13 %)

Number of brain metastases

1

112 (29 %)

33 (34 %)

16 (21 %)

37 (31 %)

26 (29 %)

0.93

2

78 (20 %)

20 (20 %)

16 (21 %)

26 (22 %)

16 (18 %)

3

48 (13 %)

9 (9 %)

11 (14 %)

15 (13 %)

13 (14 %)

 

4

53 (14 %)

10 (10 %)

13 (17 %)

16 (13 %)

14 (16 %)

 

5

24 (6 %)

8 (8 %)

4 (5 %)

6 (5 %)

6 (7 %)

 

>5

68 (18 %)

18 (18 %)

16 (21 %)

19 (16 %)

15 (17 %)

Extracranial metastases

Absent

135 (35 %)

33 (34 %)

20 (26 %)

42 (35 %)

40 (44 %)

0.13

 

Present

248 (65 %)

65 (66 %)

56 (73 %)

77 (65 %)

50 (56 %)

Gender

Female

377 (98.4 %)

97 (99 %)

75 (99 %)

116 (97 %)

89 (99 %)

0.79

 

Male

6 (1.6 %)

1 (1 %)

1 (%)

3 (3 %)

1 (1 %)

Treatment

WBRT alone

129 (34 %)

28 (29 %)

26 (34 %)

41 (34 %)

34 (38 %)

0.47

 

SRS alone

101 (26 %)

30 (31 %)

25 (33 %)

29 (24 %)

17 (19 %)

 

WBRT + SRS

85 (22 %)

22 (22 %)

11 (14 %)

30 (25 %)

22 (24 %)

 

SURG + WBRT

28 (7 %)

6 (6 %)

5 (7 %)

10 (8 %)

7 (8 %)

 

SURG + SRS + WBRT

20 (5 %)

4 (4 %)

7 (9 %)

3 (3 %)

6 (7 %)

 

SURG + SRS

19 (5 %)

7 (7 %)

2 (3 %)

6 (5 %)

4 (4 %)

 

Observation

1 (0.25 %)

1 (1 %)

0 (0 %)

0 (0 %)

0 (0 %)

Overall

 

383

98

76

119

90

 

Basal: Triple negative (HER2/ER/PR-negative)

Luminal A: HER2-negative, ER/PR-positive

Luminal B: Triple positive (HER2/ER/PR-positive)

HER2: HER2-positive, ER/PR-negative

WBRT whole Brain Radiation Therapy, SRS stereotactic Radiosurgery

Assignment of tumor subtype

Tumor subtype were approximated as follows: Basal (triple negative or HER2/ER/PR negative), Luminal A (receptor positive or HER2 negative, ER/PR positive), Luminal B (triple positive or HER2/ER/PR positive), and HER2 (HER2 positive, ER/PR negative). [7].

Statistical methods

The statistical methods defining the Breast-GPA have been previously published [5], where survival time was measured from the time of first treatment for brain metastases to the date of death or last follow-up. Prognostic factors for survival were analyzed by two methods: multivariate Cox regression (MCR) and recursive partitioning analysis (RPA). Prognostic factors found to be significant by either method were weighted relative to the magnitude of their regression coefficients to define the GPA index. Several factors found to be associated with survival were incorporated into the Breast-GPA, including tumor subtype. The expanded GPA database was further analyzed to determine if the subtype correlated with the TPDBM, number of BM, survival from PD and survival from start of treatment for BM. The association of subtype with number of BM was tested by one way ANOVA, and the association of subtype with TPDBM and survival was tested by the log-rank test.

Results

Table 3 shows the TPDBM, number of brain metastases, survival from primary diagnosis and from development of brain metastases by tumor subtype and Graded Prognostic Assessment (GPA). The median survival times (MST) for Breast-GPA 0-1, 1.5–2.0, 2.5–3.0 and 3.5–4.0 are 3.4, 7.9, 15.2 and 25.3 months, respectively. This analysis shows a statistically significant difference between the four tumor subtypes in terms of TPDBM (p < 0.01), survival from PD (p < 0.01) and survival from treatment of brain metastases (p < 0.01) however no such distinction is seen regarding the number of brain metastases (p = 0.29). Patients with Basal tumors (triple negative) develop brain metastases earlier than any other subgroup, with a median time of 27.5 months; in contrast, patients with Luminal A subtype (receptor positive) take the longest time to develop brain metastases, with a median of 54.4 months, although there was considerable variability within all subtypes. As expected, survival from diagnosis is strongly influenced by subtype, but we also found that survival after developing brain metastases actually varies significantly by subtype; the median survival for Basal, Luminal A, HER2 and Luminal B, after developing brain metastases is 7.3, 10, 17.9 and 22.9 months, respectively. Both Luminal A and B patients are receptor positive, with the only difference being HER2 status, with Luminal B positive patients being HER2-positive; the time to developing brain metastases is shorter for Luminal B (54.4 vs. 47.4 months); the median survival from diagnosis is longer for Luminal B (72. vs. 90.3 months); and the median survival after developing brain metastases is also longer for Luminal B patients (22.9 vs. 10 months).
Table 3

Time from primary diagnosis to development of brain metastases, number of brain metastases and survival in breast cancer patients by tumor subtype and graded prognostic assessment (GPA)

 

N

Median months from primary DX to brain METS (IQR)

Median number of brain METS (IQR)

MST in months from primary diagnosis (95 % CI)

MST in months from brain MET treatment (95 % CI)

Basal

90

27.5 (15.7–44.8)

3 (1–4)

39.6 (31.2–44.9)

7.3 (4.9–9.5)

HER2

119

35.8 (13.4–69.2)

2 (1–4)

66.4 (44.0–86.3)

17.9 (13.4–22.9)

Luminal B

98

47.4 (26.3–70.5)

2 (1–5)

90.3 (73.3–98.4)

22.9 (16.1–29.5)

Luminal A

76

54.4 (23.0–92.6)

3 (2–5)

72.7 (60.6–100.0)

10.0 (7.4–19.5)

  

p < 0.01

p = 0.29

p < 0.01

p < 0.01

GPA 0–1

21

28.4 (18.6–58.7)

2 (2–5)

38.6 (23.7–59.8)

3.4 (1.4–4.9)

GPA 1.5–2.0

98

30.4 (15.0–59.0)

3 (1–5)

42.5 (35.1–52.3)

7.9 (5.0–9.7)

GPA 2.5–3.0

135

43.9 (21.7–79.3)

3 (1–4)

69.5 (54.3–86.3)

15.2 (10.8–18.7)

GPA 3.5–4.0

129

41.0 (19.0–65.6)

2 (1–4)

87.0 (75.6–98.2)

25.3 (20.4–30.4)

  

p = 0.24

p = 0.13

p < 0.01

p < 0.01

Total

383

37.5 (18.6–69.2)

3 (1–4)

65.1 (56.7–75.0)

14.4 (12.0–16.3)

Discussion

Tumor subtype is an important prognostic factor for survival in patients with breast cancer and BM. The Breast-GPA provides a user-friendly prognostic index with clear separation between groups. This index can guide clinical decision-making and also be applied in the design of clinical trials. An ongoing trial (RTOG 1119) is investigating WBRT +/- lapatinib in patients with HER2-positive breast cancer and brain metastases. That trial is stratified by the Breast-GPA. Limitations of this study include the retrospective nature of the database and the long period of time during which the standards of care, including the techniques used to measure HER2 status, have evolved. There is selection bias in any retrospective study including this one. Nonetheless, these data reflect patterns of care for patients with brain metastases over the past quarter century.

In this study, we found that tumor subtype correlates with TPDBM and survival from both PD and from treatment of BM, but does not correlate with number of BM. The number of brain metastases is not a significant prognostic factor for survival in this analysis and thus is not part of the Breast-GPA. Patients should not be denied treatment because of the number of brain metastases and treatment decisions should be individualized accordingly.

The TPDBM is not a statistically significant independent prognostic factor for survival (and thus is not part of the Breast-GPA), however the TPDBM does correlate with tumor subtype. Patients with Basal and HER2 tumor subtypes had the shortest TPDBM. Patients with Basal subtype also had the shortest survival. Interestingly, patients with Luminal A breast cancer have the longest TPDBM but only 10 month median survival thereafter. HER2 positivity alone, however, does not correlate with shorter TPDBM because Basal (HER2/ER/PR negative) patients have the shortest TPDBM (27.5 months) and Luminal A (HER2-negative, ER/PR-positive) has the longest TPDBM (54.4 months). These data suggest a possible biological interaction between HER2 and ER/PR as reflected by the TPDBM and are consistent with data from other groups indicating a stronger effect of ER/PR status on the incidence of brain metastases as site of first recurrence compared to first and subsequent metastatic sites in patients with HER2-positive breast cancer [16]. Also of interest is that after the diagnosis of BM, the HER2-negative patients (Basal and Luminal A subtypes) have the shortest survival (7.3 and 10.0 months, respectively). This may reflect a treatment effect in that the HER2-positive patients (HER2 and Luminal B subtypes) receive trastuzumab, which may contribute to their longer survival due to improved systemic control, and conversely, that effect is absent in the HER2-negative patients. Another possible explanation for the shortened survival in the HER2-negative patients is that the brain metastases are genetically different from the primary tumor or systemic metastases and thus may have different response characteristics and sensitivities to treatment.

The interaction of HER2 and ER/PR status is the focus of ongoing research. Prospective studies are needed to determine if screening brain MRIs would be beneficial in patients with breast cancer. Given the clear difference in the distribution of brain metastases over time, and the stark differences in survival after BM, such studies should at a minimum take into account tumor subtype in the analysis and potentially consider subtype-specific stratification, schedules and endpoints in their design.

Notes

Acknowledgments

Dr. Mehta has served as a consultant to Abbott, Bristol-Meyers-Squibb, Elekta, Genentech, Merck, Novartis, Novocure, Tomotherapy and Viewray; he serves on the Board of Directors of Pharmacyclics, He holds stock options in Pharmacyclics and Accuray. Dr. Lin has served as a consultant to Novartis (<$10 K) and GlaxoSmithKline (<$10 K). Grant Support This research was supported in part by Grant W81XWH-062-0033 from the U.S. Department of Defense Breast Cancer Research Program, to RJW, and by NIH Grant P30- CA77598 utilizing the services of the Biostatistics Core, Masonic Cancer Center, University of Minnesota shared resource.

Conflict of interest

The authors have no conflict of interests.

References

  1. 1.
    Siegel R, Naishadham D, Jemal A (2012) Cancer statistics 2012. CA Cancer J Clin 62:10–29PubMedCrossRefGoogle Scholar
  2. 2.
    Lin NU, Bellon JR, Winer EP (2004) CNS metastases in breast cancer. J Clin Oncol 22:3608–3617PubMedCrossRefGoogle Scholar
  3. 3.
    Gaspar L, Scott C, Rotman M et al (1997) Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 37:745–751PubMedCrossRefGoogle Scholar
  4. 4.
    Sperduto PW, Kased N, Roberge D et al (2012) Summary report on the graded prognostic assessment: an accurate and facile diagnosis-specific tool to estimate survival for patients with brain metastases. J Clin Oncol 30:419–425PubMedCrossRefGoogle Scholar
  5. 5.
    Sperduto PW, Kased N, Roberge D et al (2012) Effect of tumor subtype on survival and the graded prognostic assessment for patients with breast cancer and brain metastases. Int J Radiat Oncol Biol Phys 82:2111–2117PubMedCrossRefGoogle Scholar
  6. 6.
    Gabos Z, Sinha R, Hanson J et al (2006) Prognostic significance of human epidermal growth factor receptor positivity for the development of brain metastasis after newly diagnosed breast cancer. J Clin Oncol 24:5658–5663PubMedCrossRefGoogle Scholar
  7. 7.
    Carey LA, Perou CM (2009) Gene arrays, prognosis and therapeutic interventions. In: Harris JR, Lippman ME, Morrow M, Osborne CK, editors. Diseases of the Breast. 4th edn. Philadelphia, PA, Wolters Kluwer/Lippincott Williams & Wilkins, p. 458–472Google Scholar
  8. 8.
    Perou CM, Sorlie T, Eisen MB et al (2000) Molecular portraits of human breast tumours. Nature 406:747–752PubMedCrossRefGoogle Scholar
  9. 9.
    Sorlie T, Perou CM, Tibshirani R et al (2001) Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci USA 98:10869–10874PubMedCrossRefGoogle Scholar
  10. 10.
    Nguyen PL, Taghian AG, Katz MS et al (2008) Breast cancer subtype approximated by estrogen receptor, progesterone receptor, and HER2 is associated with local and distant recurrence after breast conserving therapy. J Clin Oncol 26:2373–2378PubMedCrossRefGoogle Scholar
  11. 11.
    Andrews DW, Scott CB, Sperduto PW et al (2004) Whole brain radiation therapy with or without stereotactic radiosurgery boost for patients with one to three brain metastases: phase III results of the RTOG 9508 randomised trial. Lancet 363:1665–1672PubMedCrossRefGoogle Scholar
  12. 12.
    Sperduto PW, Wang M, Robins HI, et al. A phase III trial of whole brain radiation therapy (WBRT) and stereotactic radiosurgery (SRS) alone versus WBRT & SRS with temozolamide or erlotinib for non-small cell lung cancer and 1-3 brain metastases: Radiation Therapy Oncology Group 0320. Int J Radiat Oncol Biol Phys. Available at http://dx.doi/10.1016/j.ijrobp.2012.11.042
  13. 13.
    Le Scodan R, Jouanneau L, Massard C (2011) Brain metastases from breast cancer: prognostic significance of HER-2 overexpression, effect of trastuzumab and cause of death. BMC Cancer 11:395PubMedCrossRefGoogle Scholar
  14. 14.
    Sperduto PW, Chao ST, Sneed PK et al (2010) Diagnosis-specific prognostic factors, indexes, and treatment outcomes for patients with newly diagnosed brain metastases: A multi-institutional analysis of 4,259 patients. Int J Radiat Oncol Biol Phys 77:655–661PubMedCrossRefGoogle Scholar
  15. 15.
    Sperduto PW, Berkey B, Gaspar LE et al (2008) A new prognostic index and comparison to three other indices for patients with brain metastases: an analysis of 1,960 patients in the RTOG database. Int J Radiat Oncol Biol Phys 70:510–514PubMedCrossRefGoogle Scholar
  16. 16.
    Vas-Luis I, Ottesen RA, Hughes ME, et al (2012) Impact of hormone receptor status on patterns of recurrence and clinical outcomes among patients with human epdermal growth factor-2-positive breast cancer in the National Comprehensive Cancer Network: a prospective cohort study. Breast Cancer Res 14:R129Google Scholar

Copyright information

© Springer Science+Business Media New York 2013

Authors and Affiliations

  • Paul W. Sperduto
    • 1
    • 19
  • Norbert Kased
    • 2
  • David Roberge
    • 3
  • Samuel T. Chao
    • 4
  • Ryan Shanley
    • 5
  • Xianghua Luo
    • 5
    • 6
  • Penny K. Sneed
    • 2
    • 7
  • John Suh
    • 4
  • Robert J. Weil
    • 7
  • Ashley W. Jensen
    • 8
  • Paul D. Brown
    • 9
  • Helen A. Shih
    • 10
  • John Kirkpatrick
    • 11
  • Laurie E. Gaspar
    • 12
  • John B. Fiveash
    • 13
  • Veronica Chiang
    • 14
  • Jonathan P.S. Knisely
    • 15
  • Christina Maria Sperduto
    • 16
  • Nancy Lin
    • 17
  • Minesh Mehta
    • 18
  1. 1.Minneapolis Radiation Oncology, University of Minnesota Gamma KnifeMinneapolisUSA
  2. 2.Department of Radiation OncologyUniversity of California San FranciscoSan FranciscoUSA
  3. 3.Radiation OncologyNotre Dame Hospital, University of MontrealMontrealCanada
  4. 4.Department of Radiation OncologyCleveland ClinicClevelandUSA
  5. 5.Masonic Cancer Center, University of MinnesotaMinneapolisUSA
  6. 6.Division of BiostatisticsSchool of Public Health, University of MinnesotaMinneapolisUSA
  7. 7.Department of NeurosurgeryCleveland ClinicClevelandUSA
  8. 8.Department of Radiation OncologySanford Roger Maris Cancer CenterFargoUSA
  9. 9.MD Anderson Cancer CenterHoustonUSA
  10. 10.Harvard Medical School, Massachusetts General HospitalBostonUSA
  11. 11.Department of Radiation OncologyDuke University Medical CenterDurhamUSA
  12. 12.Department of Radiation OncologyUniversity of Colorado School of MedicineAuroraUSA
  13. 13.Radiation OncologyUniversity of Alabama Medical Center at BirminghamBirminghamUSA
  14. 14.Department of NeurosurgeryYale University School of Medicine and Yale Cancer CenterNew HavenUSA
  15. 15.Department of Radiation MedicineHofstra University School of Medicine and North Shore-Long Island Jewish Health SystemManhassetUSA
  16. 16.Dartmouth CollegeHanoverUSA
  17. 17.Department of Medical OncologyDana Farber Cancer InstituteBostonUSA
  18. 18.University of Maryland School of MedicineBaltimoreUSA
  19. 19.WaconiaUSA

Personalised recommendations