Cardiotoxicity in Patients Receiving Trastuzumab (Herceptin):
`Primary Toxicity, Synergistic or Sequential Stress, or
`Surveillance Artifact?
`
`Michael S. Ewer, Harry R. Gibbs, Joseph Swafford, and Robert S. Benjamin
`
`Human epidermal growth factor receptor-2 (HER2) is
`a member of the epidermal growth fador receptor
`family, which produces factors that are considered to
`be important mediators of cell growth. Overexpression
`of HER2, which occurs in approximately 25% to 30% of
`human breast cancers, has fostered considerable inter(cid:173)
`est in innovative therapeutic modalities designed to
`target tumor cells demonstrating such overexpression.
`Trastuzumab (Herceptinó Genentech, San Francisco,
`CA), a humanized monoclonal antibody developed to
`target the HER2 receptor, is the most widely studied
`example of such a modality. In early clinical studies
`with trastuzumab, cardiomyopathy was observed with
`a clinical expression similar to that seen with the an(cid:173)
`thracyclines (ie, a potentlally progressive decrease in
`cardiac systolic function). A number of possible expla(cid:173)
`nations for this cardiotoxicity are explored in this re(cid:173)
`port. The first is that trastuzumab has inherent toxic(cid:173)
`ity. This consideration has some theoretical interest,
`since fetal myocardial cells exhibit HER2 receptors and
`the adult myocardium expresses HER3 receptors. A
`second possibility is that sequential stresses following
`doxorubicin administration contribute to cardiac dys(cid:173)
`function. A thlrd explanation is that observational ar(cid:173)
`tlfacts lead to an overestimation of trastuzumab car(cid:173)
`diotoxicity. Approaches for additional study of the
`extent and severity of trastuzumab cardiotoxicity are
`briefly addressed.
`Semin Oncol 26 (suppl 12):96-101. Copyright © 1999 by
`W.B. Saunders Company.
`
`H" UMAN EPIDERMAL growth factor recep(cid:173)
`
`tor-2 (HER2), also known as c-erbB-2 or
`HER2/neu, is a member of the epidermal growth
`factor receptor family of receptor tyrosine kinases,
`which are considered ro be important mediators of
`cell growth, differentiation, and surviva1. 1 Overex-
`
`From the Dej)[ll"tments al Medical Specialties and Melanoma/
`Sarcoma, Divisioll ol Medicine, The University ol Texas M.D.
`Anderson Cancer Center, Houston, TX.
`Dr Swafford has received researcil suppore from Bristol-Myers
`Sqtlibh and Ccnentech, is a p,ád consultan! for Bristal-Mycrs
`Sqttibb and RhOne-Pou/enc Rorer, and is a l1lcmber ol the speal<er's
`bureaulor Bristol-Myers Squibb, Cenentech, and Rache.
`Sponsored by Genentech, [nc (Sollt11 San Francisco, CA).
`Address reprint requests to Michael S. Ewer, MD, MPH, De(cid:173)
`jJartmene al Medical Specialties, Box 70, Tile University ol Texas
`M.D. Anderson Cancer Center, 1515 Holcombe B/vd, Houston,
`TX 77030.
`Copyright © 1999 by W.B. Saunders Company
`0093-7754/99/2604-1213$10 .00/0
`
`pression or amplification of HER2 is present in a'
`variety of human malignancies. HER2 overexpres(cid:173)
`sion is detectable in approximatcly 25% to 30% of
`human breast cancers and is most often caused by
`gene amplification. z However, overexpression of
`HER2 is not unique to breast cancer. 1 t is also
`observed in other malignancics, including those
`arising from the uterine endomctrium, pancreas,
`colon, ovaries, lllngs, stomach, salivary glands, and
`heacl and neck tlllll.ors.
`Results from a number of studies have suggested
`that breast cancers that overexpress HER2 have an
`increased growth rate of malignant ce1ls. Patients
`with such tumoro have él less favorable prognosis
`when treated with conventional chemothcrapy
`compared with patients whose tumors do not over(cid:173)
`express HER2.' While the inidal response to ther(cid:173)
`apy may be similar, a significant decrcase in the
`disease-free survival is often observed in patients
`with HER2 overexpression. This difference may be
`due in part to the relatively higher growth rates of
`tumors with HER2 overexpression.4-6 Overexpres(cid:173)
`sion also appears to be associated with resistance
`to tamoxifen. 5 As therapeutic options may be
`more limited in patients wirll HER2 overexpres(cid:173)
`siol1 than in those withollt overexpression, inno(cid:173)
`vative therapies capable of manipulating 1 IER2
`overexpression are of special intcrest to physicians
`caring for these patients.
`Sclective targcting and inhibition of HER2 us(cid:173)
`ing monoclonal antibody therapy is now under
`investigation. One such agent is trastuzumab (Her(cid:173)
`ceptin; Genentcch, San Francisco, CA), a human(cid:173)
`ized monoclonal antibody. Trastuzumab exhibits
`modest antineoplastic activity when used as a sin(cid:173)
`gle agent in patients whose breast tumors overex(cid:173)
`press 1 WR2, but appears to have far greatcr poten(cid:173)
`tial, at least in some patients, when used in
`conjunction with established chemotherapy regi(cid:173)
`mens. Anthracyclines are a mainstay of treatment
`for breast cancer. They are likely to be used in
`conjunction with trastuzumab and have becn com(cid:173)
`bined in a phase III multinational trial (see Shak
`et al, pp, 71-77). Thc issue of cardiac damage
`
`96
`
`Seminars in Oncology, Vol 26, No 4, Suppl 12 (August), 1999: pp 96- 101
`
`IMMUNOGEN 2151, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`PERSPECTIVE ON TRASTUZUMAB CARDIOTOXICITY
`
`97
`
`associated with such regimens is a matter of in(cid:173)
`creasing interest and is the focus (lf this reporto
`
`CARDIOTOXICITY OF TRASTUZUMAB
`
`Preliminary information suggests that a decrease
`in cardiac systolic fUllction is a clinical character(cid:173)
`istic of trastuzumab cal'diotoxicity. While little is
`known about the mechanism of this cardiotoxic(cid:173)
`ity, a numbel' of anecdotal obsel'vations may shed
`SOll1e light on this phenomenon. When trastu(cid:173)
`zumab alone was administered in three trials to
`groups of patients who had previously received
`standard chemotherapy, an increased incidence of
`suppressed cardiac systolic function was seenj the
`crude risle estimate at 12 months was estimated to
`be 0.03. When trastuzllmab was given together
`with an anthracycline in one of these trials, this
`risle increased significantly to 0.26. The crude risle
`for patients in the randomly assigned anthracy(cid:173)
`cline-only arm was intermediate at 0.06 (S. Shak,
`personal communication, November 1998). The
`suspected cardiotoxic reaction took the form of a
`decrease in systolic function, the same clinical
`manifestation seen with anthracyclines. The de(cid:173)
`tails of these various studies and the cardiac assess(cid:173)
`ment used lO generate these data are addressed in
`other reports in this supplement.
`A superficial overview of these data suggests
`that trastuzumab is associated with low-grade cal'(cid:173)
`diotoxicity, with significant positive additive or
`synergistic e((ects when administered concomi(cid:173)
`tantly or scquential1y with anthracyclincs. The
`true nature of the cardiotoxic effects of tl'astu(cid:173)
`zumab, as wen as the nature of the carcliotoxicity
`observed with the trastuzumab-c1oxorubicin com(cid:173)
`bination, is clearly a much more complex issue
`that wil! require considerable additional study to
`be placed in clinical perspective.
`
`Lessons From the Anthracyclines
`The decrease in carcliac function reported in the
`preliminary trastuzumab studies has be en evalu(cid:173)
`ated predominantly by estimations of systolic fune(cid:173)
`tion using nuclear 111ultigated (MUGA) cardiac
`blood pool scans. Because congestive heart failure
`and a decrease in cardíac systolic function are the
`primary clinical characteristics of anthracycline
`toxicity, it is appropriate to review anthraeycline
`cardiotoxicity and its clinical manifestations to
`gain insight into what may be a similar clinical
`scenario with trastuzumab. In tlte case of a~thra-
`
`cyclines and anthraquinolones, the eventual ele(cid:173)
`struction of myocytes is the final pathway leading
`to Gudiac dysfunction. With regard to cardiac
`changes associated with trastuzumab, the mecha(cid:173)
`nism is unknown and unexplored.
`Dnxorubicin cardiotoxicity. 1 t has been well de(cid:173)
`scribed that clinical cardiotoxicity of doxorllbicin
`is rare at low cumulative dos es but becomes in(cid:173)
`creasingly more likely as the cllll1ulative dose in(cid:173)
`creases. The original correlation oi' cardiotoxicity
`and cumulative doxorubicin dose of :::::550 mg/m2
`was reported in 1973.7 The cU1l1ulative-incidence
`curve, reported in 1979, is well known,R and sug(cid:173)
`gested that approximately 5% of patients who re(cid:173)
`ceived a cutl1ulative c10se of 550 mg/m2 would be
`expected to develop symptoms of congestive heart
`failure.
`Over the past 20 years, our group has perfonned
`a large number of clinícal evaluations ane! non in(cid:173)
`vasive cardiac stuclies in patients who have re(cid:173)
`ceived anthracyclines. These evaluations have
`been carried out for a variety of clinical indica(cid:173)
`tions, but the majority have been performed in an
`attempt to identify patients who hacl demonstrated
`premature cardiac changes at rclatively low cumu(cid:173)
`lative anthracycline doses. lt was hoped that iden(cid:173)
`tifying such patients would allow us to administer
`higher cumulative c10ses to patlents without pre(cid:173)
`mature cardiac dysfunction. This strategy has be en
`largely unsuccessful. A comparison of the original
`cumulative dose-cardiotoxicity curve, detined be(cid:173)
`fore frequent noninvasive follow-up of patients,
`with a similar curve based on more recent cases
`has not allowed a higher cumulative dose as a
`consequence of heightened surveillance. What has
`been shown is that doxorllbicin is considerably
`more toxic than had be en previously thought and
`that the cumulative dos(' associated with a 5% risk
`of congestive heart failure is approximately 400
`mg/m2.s The only clear benefit of careful scrutiny
`of a very large group of doxorubicin-trcated pa(cid:173)
`tients was the discovery of cardiotoxicity when it
`would otherwise have gone unrecognized. AI(cid:173)
`though the information gained from this surveil(cid:173)
`lance did not allow us to use higher doses of the
`c1mg, it has clearly demonstl'ated that it is prudent
`to use less. 9-11
`Noninvasive tests for evaluating cardiac dysfunc(cid:173)
`tion.
`In screening for cardiotoxicity with tests
`measuring cardiac systolic function, two adclitional
`factors became evident: the tests were imperfect
`
`IMMUNOGEN 2151, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`98
`
`EWER ET AL
`
`and the eaneer patient population being observed
`for anthraeycline-re1ated eardiotoxicity is uniquely
`unsuited for sueh evaluatíon.
`Most determinations of systolie funetion use
`either nuclear imaging bIood pool seans or echo(cid:173)
`eardiography. Both of these techniques have their
`proponents, but neither is clearly reeognized to be
`superior. When performed with great care, both
`teehniques have considerable predictive vaIuc, but
`they are not infallible; some false-positive and
`false-negativc results cannot be avoided. If the
`criteria for positivity of results are seleeted nar(cid:173)
`rowly, some patients with eardiae dysfunetion will
`not have positive results; if the critcria are seleeted
`too broadly, some patients who do not have ear(cid:173)
`diac dysfunction will be reported as having a pos(cid:173)
`itive resulto Most nuclear medicine centers would
`be pleased if their technique was sufflcient1y pre(cid:173)
`cise and their reporting criteria appropriate1y se(cid:173)
`lective so that a predictive value of 90% could be
`achieved in any given clinical setting. In centers
`that perform most of their nuclear cardiology pro(cid:173)
`cedures in patients with coronary artery disease,
`for whom rhe nuclear test offered is often a sereen(cid:173)
`ing test, false-negative results may be considered a
`greater risk for patienrs than false-positive rcsults
`beeause patients with positive test results will be
`further assessed with additional, usually invasive
`procedures. Such a bias may result in overreporting
`of false-positive versus false-negative results in an(cid:173)
`thracycline-treated patients.
`Fortunately, anthracycline-associated clinical
`cardiotoxicity is rare at low cumuIative doses. At
`eumulative doses of less than 300 mg/m2, the prob(cid:173)
`ability of a doxorubicin-treated patient having ear(cid:173)
`diotoxicity is approximately 1 %j at a 90% predic(cid:173)
`tive value, a positive test report implies that for
`each true-positive result there will be nine [alse(cid:173)
`positive results. With the likelihood of cardiotox(cid:173)
`icity associated with a cumulative doxorubicin
`dose less than 300 mg/m2 and the uncertainty of
`the test procedure, the test is tota11y useless as
`either a screening or diagnostic test for the desired
`clinical application. If the test could be improved
`to the 95% predictive value level, there would stin
`be many more false-positive than true-positive re(cid:173)
`sults, suggesting cardiotoxicity. Even at a cumula(cid:173)
`tive dos e of 450 mg/m2 , the test must be evaIuated
`with great caution.
`Another factor eontributing to the clinical,
`problems of functional tests in the evaluation of
`
`doxorubicin cardiotoxicity is more complex and
`eannot be explained by simple statistical analysis.
`The normal physiologic variation seen in any
`given patient over time is augmented in the cancer
`patient by the disease process itself and its inher(cid:173)
`ent pathologic turmoil that alters metabolism, as'
`well as by the effects of the antineoplastic treat(cid:173)
`ments administered. These effects are often further
`allgmented or diminished by the impact of sup(cid:173)
`portive care therapies. Changes in the hemoglobin
`levcl, the presenee of effusions, nutritional status,
`alterations in activity lcvel, decreases in muscle
`mass, and hormonal effects a11 tend to increase or
`decrease the ejection fraetion independent of
`heart rateo We have not yet learned how to neu(cid:173)
`tralize these effects to obtain apure therapy-de(cid:173)
`pendent change in ejection fraction that can be
`evaluated independent1y. Unfortunately, these ef(cid:173)
`fects tend to cluster as the patient's disease wors(cid:173)
`ens. Thus, the greater the need fOl' an accurate
`determination of ejeetion fraction in a cancer pa(cid:173)
`tient, the more difflcult it may be to make useful
`and accurate observations from the deceivingly
`precise numbers often generated through the use
`of noninvasive studies. Nowhere has this been as
`clearly demonstrated as in the evaluation of an(cid:173)
`thracycline eardiotoxicity.
`Cardíac biopsies. The heart has considerable
`compensatory reserves, and the results of cardiac
`function tests often remain unchanged despite car(cid:173)
`diae damage and structural alteratlon of cardiac
`muscle. Morphologic evaluation overcomes some
`of the problems of quantifying anthracycline car(cid:173)
`diotoxicity, especial1y at relatively low cumulative
`doses, where spurious false-positive results are
`likely to appear and compensatory mechanisms
`may mask subtle changes in cardiac function. Mor(cid:173)
`phologic changes in myocardial ce lis callsed by
`anthracyclines have been well described. J2 At low
`cumulative anthracycline doses, vacuo le formation
`may be noted¡ at higher cumuIative doses, myofl(cid:173)
`bril dropout and frank necroscs are seen. The
`actual grading follows a scale ranging from O to 3,
`and may include half-steps in grade. The alter(cid:173)
`ations .detected by electron microscopy are clearly
`seen and quite speciüc.
`Cardiac changes can be detected far earlier by
`biopsy than with noninvasive screening tests. Bi(cid:173)
`opsy results, especially in patients who have re(cid:173)
`ceived relatively Iow doses of cardiotoxic drugs,
`have arar greater predictive value than those of
`
`IMMUNOGEN 2151, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`PERSPECTIVE ON TRASTUZUMAB CARDIOTOXICITY
`
`99
`
`functional tests. Unlike fllnctional tests, however,
`a biopsy does not detect the effects of other factors
`that may influenee eardiac performance. While
`the biopsy may only detect a minimal change
`caused by an anthracycline, a MUGA sean may
`also det~nnine the effects of coexisting ischemic
`ane! nonanthraeycline-prodllcing myopathic con(cid:173)
`sequences, which may result in a substantial clin(cid:173)
`ical effect. When a biopsy is used to make clinical
`deeisions, the information from the biopsy is often
`lIsed together with the reslllts of funetional tests
`and clinical observations. 9
`The primary role of a eardiac hiopsy is to com(cid:173)
`pare the relative toxicity of various anthracyclines
`and cardioproteetive strategies. The biopsy allows
`sueh eomparisons to be made at lower cumulative
`dos es than would otherwise be possible and with a
`eonsiderably smaller sample size. Cardiac biopsy is
`also useflll in seleeted clinical situations in which
`the etiology of eardiae dysfunction is in question,
`and the decision to stop or eontinue cardiotoxic
`ehemotherapy is crucial.
`Cardiac biopsies are clearly invasive as well as
`expensive. In experienced hands, however, they
`are safe: no deaths from cardiae biopsy were re(cid:173)
`ported in a series of 1,350 conseeutive studics
`performed in patients with eaneer.1 3 Cardiae bi(cid:173)
`opsy offers a unique opportunity to explore cardiae
`damage in an objective, focused way at a much
`earlier phase of treatment than is possible with
`other techniques.
`
`Trastuzumab Cardiotoxicity
`Detailed and conclusive studies of trastuzumab
`cardiac effeets have not been carried out. Prelim(cid:173)
`inary trials have identified a number of patients
`who have experienced eardiotoxicity associated
`with administration of rhe drug. In a phase III
`eomparative trial of patients with metastatic breast
`eancer who were treated with either ehemother(cid:173)
`apy alone or chemotherapy plus trastuzumab, car(cid:173)
`diae dysfunction was noted in 6% of patients in
`the doxorubicin-alone arm and in 27% of patients
`in the doxorubicin plus trastuzumab armo For the
`purposes of this study, cardiotoxicity was defined as
`(1) eardiomyopathy, eharaeterized by a deerease in
`cardiae ejeetion fraction associated with abnormal
`myoeardial wall motion that was either global or
`more severe in the septum¡ (2) symptoms of con(cid:173)
`gestive heart failure; (3) associated signs of con(cid:173)
`gestive heart failure including but not limited' to a
`
`diastolic gallop and/or taehycardia; or (4) a decline
`in eardiac ejection fraetion (ram baseline of at
`least S percentage poin!s to below S 5% with signs
`and symptoms or a deerease in eardiac ejection
`fraction of at least 10 percentage points to below
`55% without signs and symptoms (S. Shak, per(cid:173)
`sonal communication, November 1998).
`A number of hypotheses regarding the observed
`toxicity can be considered. The first is that trastu(cid:173)
`zumab has inherent toxicity. Thus, like the an(cid:173)
`thracyclines, it is independently cardiotoxic. The
`second hypothesis is tha! the drug amplifies the
`cardiotoxic effeet or has an additive or expressive
`toxic effect when given together or sequentially
`with anthraeyclines or other agents that have in(cid:173)
`herent toxicity. A third hypothesis is that the
`described toxicity of trastuzumab represents sorne
`fonn of observational artifact. A combinatíon of
`these possibilities (¡e, additive effects plus inherent
`toxicity) may also occur. The three possible expla(cid:173)
`nations will he considered in their pure formo
`Inherent
`trastllzumab cardiac
`tm .. icity. Larger
`doses of the drug, longer periods of observation, or
`more sensitive cardiac testing will be needed to
`demonstrate, confirm, and quantify any inherent
`trastuzumab toxieity. The mechanism underlying
`the reported cardiac changes, if they in fact exist,
`is unknown. There have been no reported evalu(cid:173)
`ations of cardiae biopsy in patients who have re(cid:173)
`ceived trastuzumab without other agents; only in(cid:173)
`freqllent decreases in the ejection fraetions of
`these patients have been reported. Based on the
`limited data at hand, it seems unlikely that inher(cid:173)
`ent cardiac toxicity will be a significant limiting
`factor in the clinical use of trastuzumab.
`Toxic sequelae related to additive or sequential ef(cid:173)
`fects of trastuzumab with anthracyclines or other
`agents exhibiting inherent toxicity. The natural his(cid:173)
`tory of anthracycline cardiotoxicity is complexo A
`simple explanatíon of the variation among pa(cid:173)
`tients in susceptibility to toxic reaetions i8 not
`available, and mueh is yet to be learned as to why
`some patients experienee cardiotoxicity sooner
`than others following exposure to the anthracy(cid:173)
`cline. One explanation is that sequential stresses
`playa role in the clinical speetrum of toxicity.14
`Our group has hypothesized that sllbclinical dam(cid:173)
`age occurs on or shortly after exposure to the toxic
`agent. However, while apparent at an ultrastrue(cid:173)
`tural level, this damage is not easily detectable
`with conventional methods of studying cardiac
`
`IMMUNOGEN 2151, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`100
`
`EWER ET AL
`
`changes in the clinical setting. Cardiac reserves
`and compensatory mechanisms may mask the
`damage until the ability to compensate Qecomes
`compromised. It is possible
`that the cardiac
`changes noted with trastuzumab may represent
`some form of. sequential stress on a previously
`altered myocardium or an inhibition of an ongoing
`repair mechanism. In the studies noted aboye, the
`grcat majority of patients who experieneed
`changes in cardiae parameters following trastu(cid:173)
`zumab administration had been previously exposed
`to doxoruhicin. The sequential stresses or inhib(cid:173)
`ited repair may not neeessarily be caused by the
`administration of trastuzumab, hut may be related
`to one or a combination of several secondary met(cid:173)
`abolic effects seen following such administration
`(eg, electrolyte shifts or fluctuations in cil'culating
`bio logic mediators).
`Observational artifacts as an explanation for tras(cid:173)
`tuzumab cardiotoxicity. The pitfalls regarding the
`recognition of a possible eardiotoxic effect in pa(cid:173)
`tients who have received prior cardiotoxic medi(cid:173)
`cation with a widely variable express ion were al(cid:173)
`luded to ahoye. Moreover, the underlying disease
`often mimics the very signs and symptoms used to
`define the nature of eardiotoxicity when studied by
`methods of suboptimal predictive value, discussed
`aboye. Clearly, focused surveillance will increase
`the likclihood of finding signs and symptoms of
`cardiotoxicity in any study population. This form
`of study artifact must be eonsidered in any objec(cid:173)
`ti ve overview of the rclationship between tras tu(cid:173)
`zumab and the clinical reeognition of cardiotoxic(cid:173)
`ity.
`
`SUMMARY ANO CONCLUSIONS
`
`This report has discussed the possible nature of
`cardiotoxicity associated with trastuzumab. The
`task of placing this new, potentially important
`agent into perspective would have been facilitated
`and the end-product would have been more useful
`had the drug been uniformly studied and cardiac
`changes been analyzed by methods with better
`predictive value. This luxury, howevel', is not
`available. Nevertheless, we can apply some of
`what we know about the observed cardiac changes
`in patients who received trastuzumab and can offer
`some suggestions for flltllre paths to explore.
`If trastuzumab demonstl'ates inhercnt toxicity, it
`will prabably be best detected in clinical studies
`using cardiac biopsies evaluated by a pathologist
`
`who is highly trained in recognizing and grading
`cm{iac changes. It is possiblc that the morpho(cid:173)
`logic changes caused by trastllzumab may be dif,
`ferent fram those caused by anthracyclines; this
`possibility should be actively investigated. As with
`anthracyclines, a combination of functional and
`morphologic parameters may be more llseful than
`either modality alone for detecting cardiotoxicity.
`Sequential stresses caused by trastuzumah and
`observational artifacts of the clinical data will be
`much more difficult to evaluate than inherent
`toxicity. Seqllential stl'esses involve two 01' more
`variables with very diffcrent biological and clinical
`express ion that oceur simultaneouslYi detection of
`observational artifacts wil! require large, blinded
`studies. Again, a combinatían of biopsy and fune(cid:173)
`tional determinations may offer the best chance of
`gaining insight into these problems.
`A better understanding of trastuzumah cardio,
`toxicity ís essential, as a largc numbcr of patients
`who received prior treatment with anthracyclines
`are likely to benefit from this interesting new
`agent. The trllc risk of toxicity must be defined if
`our patients are to enjoy maximal hendit fram
`these new treatment strategies Witl:lOut undue iat(cid:173)
`rogenic complications.
`
`REFERENCES
`
`1. Slamon D], Clark GM, Wong SG, el al: Human breast
`caneer: Correlation of relapse and survival \Vith amplification
`o( the HER-2/neu oncogene. Scicnce 235:177-182, 1987
`2. Berger MS, Loeher GW, Saurcr S, et al: Conelation of
`c-erbB· 2 gene amplification and prole in exprcssion in human
`breast carcinoma \Vith nodal status and nuclear grading. Caneer
`Res 48:123il-1243, 1988
`3. Press MF, Pike MC, Chazin VR, et al: Her-2/neu expres(cid:173)
`sion in node-negative breast caneer: l)ireet tissue quantitation
`by computerízed image analysis and association oí overexpres(cid:173)
`sion \Virh increased risk of reeurrent disease. Cancer Res 53:
`4950-4970, 1993
`4. Slamon DJ, Godolphin W, Jones LA, et al: Studies of the
`HER·2/neu protooncogene in human breast and ovarían can(cid:173)
`ceL Science 244:707-712, 1989
`5. Borg A, Baldetorp B, Fernéi M, et al: ERBB2 amplification
`is associated \Vith tamoxifcn resistanee in steroid-receptor pos(cid:173)
`itive breast cancer. Caneer Len 81:137-144, 1994
`6. Narita M, Nakao K, Ogino N, et al: Independent prog(cid:173)
`nostlc factors in breast cancer patients. Am J Surg 175:73·75,
`1998
`7. Lefrak E, Pitha], Rosenheim S, et al: A c1inicopathologic
`analysis of Adriamycin cardiotoxicity. Cancer 32:302-314,
`1973
`8. Yon Hoff DD, Layard MW, Basa P, et al: Risk factors for
`doxorubicin-induced congestive heart failure. Ann Intern Med
`91:710-717,1979
`
`IMMUNOGEN 2151, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`PERSPECTIVE ON TRASTUZUMAB CARDIOTOXICITY
`
`101
`
`9. Ewer MS, Benjamin RS: Cardiac complicat lons, in
`Holland JF, Frei E 1lI, Bast jR, et al (eds): Caneer Medicine
`(ed 4). Baltimore, MD, Williams & Wilkins, 1997, pp 3197-
`3215
`10. Swain SM, Whaley FS, Ger-ber MC, et al: Delayed
`administration of dexrazoxane pro'vides eardioprotection for
`patients with advanced breast caneer treated Wilh doxorubicin(cid:173)
`containing therapy. j Clin OncoI15:1333-1340, 1997
`11. Gianni L, Vigano L, Locatelli A, et al: Human pharrna(cid:173)
`cokinetic characterization and in vitro study of the interaction
`
`between doxorubicin and r"clitaxel in patients with brcast
`cancer. J Clin OncoI15:1906-1915, 1997
`12. Mackay B, Keyes LM, Benjamín RS, et al: Cardiac
`hiopsy. Tex Soc Electron Microscopy 11:7-15, 1981
`13. Ewer M, Carrasco CH, Mackay B, et al: Cardiac hiopsy
`procedures at a caneer center. Proc Am Soc Clin Oncol 10:
`136a, 1991 (abstr)
`14. Ali MK, Ewer MS, Gibbs HR, et al: Late doxorubicin(cid:173)
`associated cardiotoxicity in childrcn: The possible role nf in(cid:173)
`tcrcurrent viral infection. Cancer 74:182-188, 1994
`
`IMMUNOGEN 2151, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676