The
n e w e ng l a n d j o u r na l
gone bariatric surgery, particularly because of the phenomenon of the dumping syndrome.1 It is true that the dumping syndrome can be seen in women undergoing testing for gestational diabetes. In one recent study, 11 of 19 women (58%) had hypoglycemia in response to the glucose load, which could affect test results.2 Although van Beek and Lutgers discuss the use of fasting glucose levels and continuous glucose monitoring to screen for gestational diabetes, we do not have much guidance for how such tools would be used. In particular, fasting blood glucose levels have been shown to have poor sensitivity for the diagnosis of gestational diabetes.3 The use of self-monitoring of blood glucose levels to diagnose gestational diabetes may be the best approach for women who have had bariatric surgery until we have better data on other methods.4 In the end, these letter writers make the point that we need more research — basic, clinical,
of
m e dic i n e
and translational — to better understand what is underlying the effects of bariatric surgery, particularly in women who are or would like to become pregnant. Aaron B. Caughey, M.D., Ph.D. Oregon Health and Science University Portland, OR Since publication of his article, the author reports no further potential conflict of interest. 1. Tack J, Arts J, Caenepeel P, De Wulf D, Bisschops R. Patho-
physiology, diagnosis and management of postoperative dumping syndrome. Nat Rev Gastroenterol Hepatol 2009;6:583-90. 2. Freitas C, Araújo C, Caldas R, Lopes DS, Nora M, Monteiro MP. Effect of new criteria on the diagnosis of gestational diabetes in women submitted to gastric bypass. Surg Obes Relat Dis 2014;10:1041-6. 3. Anderson V, Ye C, Sermer M, et al. Fasting capillary glucose as a screening test for ruling out gestational diabetes mellitus. J Obstet Gynaecol Can 2013;35:515-22. 4. Allard C, Sahyouni E, Menard J, et al. Gestational diabetes mellitus identification based on self-monitoring of blood glucose. Can J Diabetes 2015;39:162-8. DOI: 10.1056/NEJMc1503863
Ovarian Protection during Adjuvant Chemotherapy To the Editor: In the study by Moore et al. (March 5 issue),1 a statistically significant improvement in disease-free and overall survival was observed in the goserelin group versus the control group, together with a higher number of pregnancies. The authors report that this improvement was “unexpected” and perhaps due to the gonadotropin-releasing hormone (GnRH) receptors on breast-cancer cells. An alternative explanation might be proposed. It has been shown that pregnancy after breast cancer does not decrease survival. In a preplanned analysis conducted in patients with estrogen-receptor–negative breast cancer, the effect of pregnancy was even protective, with a hazard ratio for diseasefree and overall survival of 0.75 and 0.54, respectively.2 Even if the biologic mechanisms of this effect have not been completely elucidated, it is plausible that pregnancy might affect outcomes.3 Thus, the possible confounding effect of pregnancy on survival in the Prevention of Early Menopause Study (POEMS) could be of relevance. Given the small number of pregnancies, it would be interesting to know when they occurred in each group
2268
and if relapses had a different distribution in patients who became pregnant as compared with patients who did not. Fedro A. Peccatori, M.D., Ph.D. European Institute of Oncology Milan, Italy [email protected] No potential conflict of interest relevant to this letter was reported. 1. Moore HCF, Unger JM, Phillips K-A, et al. Goserelin for
ovarian protection during breast-cancer adjuvant chemotherapy. N Engl J Med 2015;372:923-32. 2. Azim HA Jr, Kroman N, Paesmans M, et al. Prognostic impact of pregnancy after breast cancer according to estrogen receptor status: a multicenter retrospective study. J Clin Oncol 2013;31:73-9. 3. Azim HA Jr, Santoro L, Pavlidis N, et al. Safety of pregnancy following breast cancer diagnosis: a meta-analysis of 14 studies. Eur J Cancer 2011;47:74-83. DOI: 10.1056/NEJMc1504241
To the Editor: We have some questions about the use of GnRH analogs to preserve fertility in women with breast cancer who are undergoing chemotherapy. First, when we recalculated the numbers on the basis of patients attempting to
n engl j med 372;23 nejm.org june 4, 2015
The New England Journal of Medicine Downloaded from nejm.org at NYU WASHINGTON SQUARE CAMPUS on June 4, 2015. For personal use only. No other uses without permission. Copyright © 2015 Massachusetts Medical Society. All rights reserved.
correspondence
become pregnant rather than all women in each group, we did not find a significant difference in the rate of pregnancy (22 of 25 patients who received a GnRH analog vs. 12 of 18 who did not; Yates’s X2 = 1.73; P = 0.188). Second, 6 months of amenorrhea is an imperfect primary end point and surrogate for ovarian failure, since perimenopausal women can still have irregular menstruation. Third, without the use of placebo or blinding, women receiving the analog could interpret any bleeding as menstruation. Fourth, since levels of follicle-stimulating hormone (FSH), estradiol, and inhibin-B fluctuate throughout the menstrual cycle, random measurements cannot be relied on as indicators of ovarian reserve. Randomized trials using a less cycle-dependent marker (antiMüllerian hormone) or equally reliable antral follicle counts did not report a benefit with a GnRH analog.1-3 In addition, the presumed target of the analog, the primordial follicles that make up the ovarian reserve, are not hormone-sensitive.4 While the debate continues, the safer course may be to refrain from relying on ovarian suppression by a GnRH analog to preserve fertility at the expense of the oocyte or embryo cryopreservation. Kutluk Oktay, M.D. New York Medical College New York, NY [email protected]
Kenny Rodriguez-Wallberg, M.D., Ph.D. Karolinska University Hospital Stockholm, Sweden
Pamela Munster, M.D. University of California San Francisco, CA Dr. Oktay reports serving as cochair of the committee for guidelines on fertility preservation for the American Society of Clinical Oncology. No other potential conflict of interest relevant to this letter was reported. 1. Gerber B, von Minckwitz G, Stehle H, et al. Effect of lutein-
izing hormone-releasing hormone agonist on ovarian function after modern adjuvant breast cancer chemotherapy: the GBG 37 ZORO study. J Clin Oncol 2011;29:2334-41. 2. Munster PN, Moore AP, Ismail-Khan R, et al. Randomized trial using gonadotropin-releasing hormone agonist triptorelin for the preservation of ovarian function during (neo)adjuvant chemotherapy for breast cancer. J Clin Oncol 2012;30:533-8. 3. Elgindy EA, El-Haieg DO, Khorshid OM, et al. Gonadatrophin suppression to prevent chemotherapy-induced ovarian damage: a randomized controlled trial. Obstet Gynecol 2013; 121:78-86. 4. Oktay K, Sönmezer M. Gonadotropin-releasing hormone analogs in fertility preservation-lack of biological basis? Nat Clin Pract Endocrinol Metab 2008;4:488-9. DOI: 10.1056/NEJMc1504241
The Authors Reply: The point raised by Peccatori was discussed by the authors during the preparation of the POEMS manuscript. In a retrospective study by Azim et al., pregnancy after a breast-cancer diagnosis was associated with significantly improved overall survival but not disease-free survival.1 Women who become pregnant may have better-than-average health and may therefore also be more likely to have prolonged survival. The design of POEMS does not enable us to separate the relative well-being that may lead to both a pregnancy outcome and longer survival as opposed to a causal association. Oktay and colleagues raise other points. We caution against using the number of patients attempting pregnancy as a denominator when comparing pregnancy rates. Multiple biases are possi ble. Attempted pregnancy or a report of attempted pregnancy could be influenced by the intervention assignment. Furthermore, some pregnancies occurred in women who did not report an attempt to become pregnant. Given these considerations, the most valid denominator for comparing pregnancy outcomes is the entire randomized population that can be evaluated, which is what we used in the study. We agree that amenorrhea is an imperfect surrogate for ovarian failure. However, recovery of normal menses is uncommon among breastcancer patients who have amenorrhea 2 years after chemotherapy.2 Although it is unlikely that the absence of blinding to treatment group would have a meaningful influence on a woman’s ability to accurately record whether or not she had menstrual bleeding during the study observation period, it is important to note that we also incorporated FSH level, an objective measure, in the primary end point. Taken together, the composite end point is a more robust measure of ovarian failure. Infertility is one important consequence of early menopause, but we stress that POEMS was designed to evaluate the effect of a GnRH analog on ovarian failure rates, not ovarian reserve. Secondary end points that involved assessment of pregnancy outcomes strongly suggest that the prevention of ovarian failure improves fertility prospects. Giving a patient a GnRH analog during chemotherapy does not preclude the use of the traditional techniques used in assisted reproduction. Over and above the fertility issue, the
n engl j med 372;23 nejm.org june 4, 2015
2269
The New England Journal of Medicine Downloaded from nejm.org at NYU WASHINGTON SQUARE CAMPUS on June 4, 2015. For personal use only. No other uses without permission. Copyright © 2015 Massachusetts Medical Society. All rights reserved.
The
n e w e ng l a n d j o u r na l
POEMS approach reduces the onset of premature menopause and its distressing consequences. Halle C.F. Moore, M.D. Cleveland Clinic Taussig Cancer Institute Cleveland, OH [email protected]
Joseph M. Unger, Ph.D. Fred Hutchinson Cancer Research Center Seattle, WA
Kathy S. Albain, M.D.
of
m e dic i n e
Since publication of their article, the authors report no further potential conflict of interest. 1. Azim HA Jr, Kroman N, Paesmans M, et al. Prognostic im-
pact of pregnancy after breast cancer according to estrogen receptor status: a multicenter retrospective study. J Clin Oncol 2013;31:73-9. 2. Sukumvanich P, Case LD, Van Zee K, et al. Incidence and time course of bleeding after long-term amenorrhea after breast cancer treatment: a prospective study. Cancer 2010;116:3102-11. DOI: 10.1056/NEJMc1504241
Loyola University Chicago Medical Center Maywood, IL
Bortezomib-Based Therapy for Mantle-Cell Lymphoma To the Editor: Robak et al. (March 5 issue)1 reported the outcome of a study of the treatment of mantle-cell lymphoma in which 487 patients were randomly assigned to receive R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) or VR-CAP (R-CHOP regimen, but with bortezomib substituted for vincristine). The difference in progression-free survival was significant, favoring VR-CAP, whereas the difference in overall survival was not significant. The curves for progression-free survival (Fig. 1 of the article) diverged at 1 year, then connected again after 4 years. The curves for overall survival began to diverge after 1 year and did not join at a later time. Since in the calculation of progression-free survival, the events are disease progression or death, what is the explanation for the different shapes of the curves? Were there more late relapses in the VR-CAP group? If so, the overall survival curves would ultimately be expected to join rather than diverge. Moreover, in an elderly patient population (median age, 65 to 66 years), it would be of interest to know how many patients died from lymphoma, from therapy-related complications, and from other causes. Michele Baccarani, M.D.
To the Editor: Robak et al. have shown that substituting bortezomib for vincristine in R-CHOP results in an improvement in median progression-free survival in patients with newly diagnosed mantle-cell lymphoma. However, we have a few questions regarding the study results and their applicability. The median progression-free survival was 14.4 months in the R-CHOP group, which is much lower than that reported by Rummel et al.1 (40.9 months with R-CHOP). Similarly, the time to treatment failure (a more stringent end point) of 21 months was earlier than that in another randomized trial in which the R-CHOP regimen was compared with the CHOP regimen.2 The frequency of computed tomographic (CT) scanning (at intervals of 6 to 8 weeks) alone cannot explain the difference in progression-free survival between different studies in which the same therapy was provided, as suggested by the authors. The benefits of the VR-CAP regimen come at the price of remarkable hematologic toxicity. Bendamustine plus rituximab would appear to be a safer option in a similar patient population, even though a head-to-head comparison is not available.
University of Bologna Bologna, Italy [email protected] No potential conflict of interest relevant to this letter was reported.
Manju Sengar M.D., D.M. Hasmukh Jain, M.D., D.M.
1. Robak T, Huang H, Jin J, et al. Bortezomib-based therapy for
newly diagnosed mantle-cell lymphoma. N Engl J Med 2015;372: 944-53.
Tata Memorial Centre Mumbai, India [email protected] No potential conflict of interest relevant to this letter was reported.
DOI: 10.1056/NEJMc1504240
2270
n engl j med 372;23 nejm.org june 4, 2015
The New England Journal of Medicine Downloaded from nejm.org at NYU WASHINGTON SQUARE CAMPUS on June 4, 2015. For personal use only. No other uses without permission. Copyright © 2015 Massachusetts Medical Society. All rights reserved.
n e w e ng l a n d j o u r na l
gone bariatric surgery, particularly because of the phenomenon of the dumping syndrome.1 It is true that the dumping syndrome can be seen in women undergoing testing for gestational diabetes. In one recent study, 11 of 19 women (58%) had hypoglycemia in response to the glucose load, which could affect test results.2 Although van Beek and Lutgers discuss the use of fasting glucose levels and continuous glucose monitoring to screen for gestational diabetes, we do not have much guidance for how such tools would be used. In particular, fasting blood glucose levels have been shown to have poor sensitivity for the diagnosis of gestational diabetes.3 The use of self-monitoring of blood glucose levels to diagnose gestational diabetes may be the best approach for women who have had bariatric surgery until we have better data on other methods.4 In the end, these letter writers make the point that we need more research — basic, clinical,
of
m e dic i n e
and translational — to better understand what is underlying the effects of bariatric surgery, particularly in women who are or would like to become pregnant. Aaron B. Caughey, M.D., Ph.D. Oregon Health and Science University Portland, OR Since publication of his article, the author reports no further potential conflict of interest. 1. Tack J, Arts J, Caenepeel P, De Wulf D, Bisschops R. Patho-
physiology, diagnosis and management of postoperative dumping syndrome. Nat Rev Gastroenterol Hepatol 2009;6:583-90. 2. Freitas C, Araújo C, Caldas R, Lopes DS, Nora M, Monteiro MP. Effect of new criteria on the diagnosis of gestational diabetes in women submitted to gastric bypass. Surg Obes Relat Dis 2014;10:1041-6. 3. Anderson V, Ye C, Sermer M, et al. Fasting capillary glucose as a screening test for ruling out gestational diabetes mellitus. J Obstet Gynaecol Can 2013;35:515-22. 4. Allard C, Sahyouni E, Menard J, et al. Gestational diabetes mellitus identification based on self-monitoring of blood glucose. Can J Diabetes 2015;39:162-8. DOI: 10.1056/NEJMc1503863
Ovarian Protection during Adjuvant Chemotherapy To the Editor: In the study by Moore et al. (March 5 issue),1 a statistically significant improvement in disease-free and overall survival was observed in the goserelin group versus the control group, together with a higher number of pregnancies. The authors report that this improvement was “unexpected” and perhaps due to the gonadotropin-releasing hormone (GnRH) receptors on breast-cancer cells. An alternative explanation might be proposed. It has been shown that pregnancy after breast cancer does not decrease survival. In a preplanned analysis conducted in patients with estrogen-receptor–negative breast cancer, the effect of pregnancy was even protective, with a hazard ratio for diseasefree and overall survival of 0.75 and 0.54, respectively.2 Even if the biologic mechanisms of this effect have not been completely elucidated, it is plausible that pregnancy might affect outcomes.3 Thus, the possible confounding effect of pregnancy on survival in the Prevention of Early Menopause Study (POEMS) could be of relevance. Given the small number of pregnancies, it would be interesting to know when they occurred in each group
2268
and if relapses had a different distribution in patients who became pregnant as compared with patients who did not. Fedro A. Peccatori, M.D., Ph.D. European Institute of Oncology Milan, Italy [email protected] No potential conflict of interest relevant to this letter was reported. 1. Moore HCF, Unger JM, Phillips K-A, et al. Goserelin for
ovarian protection during breast-cancer adjuvant chemotherapy. N Engl J Med 2015;372:923-32. 2. Azim HA Jr, Kroman N, Paesmans M, et al. Prognostic impact of pregnancy after breast cancer according to estrogen receptor status: a multicenter retrospective study. J Clin Oncol 2013;31:73-9. 3. Azim HA Jr, Santoro L, Pavlidis N, et al. Safety of pregnancy following breast cancer diagnosis: a meta-analysis of 14 studies. Eur J Cancer 2011;47:74-83. DOI: 10.1056/NEJMc1504241
To the Editor: We have some questions about the use of GnRH analogs to preserve fertility in women with breast cancer who are undergoing chemotherapy. First, when we recalculated the numbers on the basis of patients attempting to
n engl j med 372;23 nejm.org june 4, 2015
The New England Journal of Medicine Downloaded from nejm.org at NYU WASHINGTON SQUARE CAMPUS on June 4, 2015. For personal use only. No other uses without permission. Copyright © 2015 Massachusetts Medical Society. All rights reserved.
correspondence
become pregnant rather than all women in each group, we did not find a significant difference in the rate of pregnancy (22 of 25 patients who received a GnRH analog vs. 12 of 18 who did not; Yates’s X2 = 1.73; P = 0.188). Second, 6 months of amenorrhea is an imperfect primary end point and surrogate for ovarian failure, since perimenopausal women can still have irregular menstruation. Third, without the use of placebo or blinding, women receiving the analog could interpret any bleeding as menstruation. Fourth, since levels of follicle-stimulating hormone (FSH), estradiol, and inhibin-B fluctuate throughout the menstrual cycle, random measurements cannot be relied on as indicators of ovarian reserve. Randomized trials using a less cycle-dependent marker (antiMüllerian hormone) or equally reliable antral follicle counts did not report a benefit with a GnRH analog.1-3 In addition, the presumed target of the analog, the primordial follicles that make up the ovarian reserve, are not hormone-sensitive.4 While the debate continues, the safer course may be to refrain from relying on ovarian suppression by a GnRH analog to preserve fertility at the expense of the oocyte or embryo cryopreservation. Kutluk Oktay, M.D. New York Medical College New York, NY [email protected]
Kenny Rodriguez-Wallberg, M.D., Ph.D. Karolinska University Hospital Stockholm, Sweden
Pamela Munster, M.D. University of California San Francisco, CA Dr. Oktay reports serving as cochair of the committee for guidelines on fertility preservation for the American Society of Clinical Oncology. No other potential conflict of interest relevant to this letter was reported. 1. Gerber B, von Minckwitz G, Stehle H, et al. Effect of lutein-
izing hormone-releasing hormone agonist on ovarian function after modern adjuvant breast cancer chemotherapy: the GBG 37 ZORO study. J Clin Oncol 2011;29:2334-41. 2. Munster PN, Moore AP, Ismail-Khan R, et al. Randomized trial using gonadotropin-releasing hormone agonist triptorelin for the preservation of ovarian function during (neo)adjuvant chemotherapy for breast cancer. J Clin Oncol 2012;30:533-8. 3. Elgindy EA, El-Haieg DO, Khorshid OM, et al. Gonadatrophin suppression to prevent chemotherapy-induced ovarian damage: a randomized controlled trial. Obstet Gynecol 2013; 121:78-86. 4. Oktay K, Sönmezer M. Gonadotropin-releasing hormone analogs in fertility preservation-lack of biological basis? Nat Clin Pract Endocrinol Metab 2008;4:488-9. DOI: 10.1056/NEJMc1504241
The Authors Reply: The point raised by Peccatori was discussed by the authors during the preparation of the POEMS manuscript. In a retrospective study by Azim et al., pregnancy after a breast-cancer diagnosis was associated with significantly improved overall survival but not disease-free survival.1 Women who become pregnant may have better-than-average health and may therefore also be more likely to have prolonged survival. The design of POEMS does not enable us to separate the relative well-being that may lead to both a pregnancy outcome and longer survival as opposed to a causal association. Oktay and colleagues raise other points. We caution against using the number of patients attempting pregnancy as a denominator when comparing pregnancy rates. Multiple biases are possi ble. Attempted pregnancy or a report of attempted pregnancy could be influenced by the intervention assignment. Furthermore, some pregnancies occurred in women who did not report an attempt to become pregnant. Given these considerations, the most valid denominator for comparing pregnancy outcomes is the entire randomized population that can be evaluated, which is what we used in the study. We agree that amenorrhea is an imperfect surrogate for ovarian failure. However, recovery of normal menses is uncommon among breastcancer patients who have amenorrhea 2 years after chemotherapy.2 Although it is unlikely that the absence of blinding to treatment group would have a meaningful influence on a woman’s ability to accurately record whether or not she had menstrual bleeding during the study observation period, it is important to note that we also incorporated FSH level, an objective measure, in the primary end point. Taken together, the composite end point is a more robust measure of ovarian failure. Infertility is one important consequence of early menopause, but we stress that POEMS was designed to evaluate the effect of a GnRH analog on ovarian failure rates, not ovarian reserve. Secondary end points that involved assessment of pregnancy outcomes strongly suggest that the prevention of ovarian failure improves fertility prospects. Giving a patient a GnRH analog during chemotherapy does not preclude the use of the traditional techniques used in assisted reproduction. Over and above the fertility issue, the
n engl j med 372;23 nejm.org june 4, 2015
2269
The New England Journal of Medicine Downloaded from nejm.org at NYU WASHINGTON SQUARE CAMPUS on June 4, 2015. For personal use only. No other uses without permission. Copyright © 2015 Massachusetts Medical Society. All rights reserved.
The
n e w e ng l a n d j o u r na l
POEMS approach reduces the onset of premature menopause and its distressing consequences. Halle C.F. Moore, M.D. Cleveland Clinic Taussig Cancer Institute Cleveland, OH [email protected]
Joseph M. Unger, Ph.D. Fred Hutchinson Cancer Research Center Seattle, WA
Kathy S. Albain, M.D.
of
m e dic i n e
Since publication of their article, the authors report no further potential conflict of interest. 1. Azim HA Jr, Kroman N, Paesmans M, et al. Prognostic im-
pact of pregnancy after breast cancer according to estrogen receptor status: a multicenter retrospective study. J Clin Oncol 2013;31:73-9. 2. Sukumvanich P, Case LD, Van Zee K, et al. Incidence and time course of bleeding after long-term amenorrhea after breast cancer treatment: a prospective study. Cancer 2010;116:3102-11. DOI: 10.1056/NEJMc1504241
Loyola University Chicago Medical Center Maywood, IL
Bortezomib-Based Therapy for Mantle-Cell Lymphoma To the Editor: Robak et al. (March 5 issue)1 reported the outcome of a study of the treatment of mantle-cell lymphoma in which 487 patients were randomly assigned to receive R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) or VR-CAP (R-CHOP regimen, but with bortezomib substituted for vincristine). The difference in progression-free survival was significant, favoring VR-CAP, whereas the difference in overall survival was not significant. The curves for progression-free survival (Fig. 1 of the article) diverged at 1 year, then connected again after 4 years. The curves for overall survival began to diverge after 1 year and did not join at a later time. Since in the calculation of progression-free survival, the events are disease progression or death, what is the explanation for the different shapes of the curves? Were there more late relapses in the VR-CAP group? If so, the overall survival curves would ultimately be expected to join rather than diverge. Moreover, in an elderly patient population (median age, 65 to 66 years), it would be of interest to know how many patients died from lymphoma, from therapy-related complications, and from other causes. Michele Baccarani, M.D.
To the Editor: Robak et al. have shown that substituting bortezomib for vincristine in R-CHOP results in an improvement in median progression-free survival in patients with newly diagnosed mantle-cell lymphoma. However, we have a few questions regarding the study results and their applicability. The median progression-free survival was 14.4 months in the R-CHOP group, which is much lower than that reported by Rummel et al.1 (40.9 months with R-CHOP). Similarly, the time to treatment failure (a more stringent end point) of 21 months was earlier than that in another randomized trial in which the R-CHOP regimen was compared with the CHOP regimen.2 The frequency of computed tomographic (CT) scanning (at intervals of 6 to 8 weeks) alone cannot explain the difference in progression-free survival between different studies in which the same therapy was provided, as suggested by the authors. The benefits of the VR-CAP regimen come at the price of remarkable hematologic toxicity. Bendamustine plus rituximab would appear to be a safer option in a similar patient population, even though a head-to-head comparison is not available.
University of Bologna Bologna, Italy [email protected] No potential conflict of interest relevant to this letter was reported.
Manju Sengar M.D., D.M. Hasmukh Jain, M.D., D.M.
1. Robak T, Huang H, Jin J, et al. Bortezomib-based therapy for
newly diagnosed mantle-cell lymphoma. N Engl J Med 2015;372: 944-53.
Tata Memorial Centre Mumbai, India [email protected] No potential conflict of interest relevant to this letter was reported.
DOI: 10.1056/NEJMc1504240
2270
n engl j med 372;23 nejm.org june 4, 2015
The New England Journal of Medicine Downloaded from nejm.org at NYU WASHINGTON SQUARE CAMPUS on June 4, 2015. For personal use only. No other uses without permission. Copyright © 2015 Massachusetts Medical Society. All rights reserved.
