Page 1
`
`Lilly Research Laboratories
`A Diviin of Eli Lilly nd Corripany
`
`LiUy Corporate Cenler
`Indianapoiie, Indiana 462ß5
`{31 7 276-2000
`
`December 3, 1999
`
`IND Safety Report Follow-Up and
`Request for FDA Input
`
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Division of Oncologic Drug Products, HFD-150
`AUn: Mr. Alvis Dunson
`1451 Rockville Pike
`Rockville, MD 20852-1448
`
`Subject: IND 4OO61, MTA (1X231514) - Serial no. 195
`Supplementation with Folic Acid and Vitamin B12 To Reduce
`Toxicity In Patients Receiving LY231 514
`
`Recently, Lilly sent a letter to investigators informing them to exclude patients
`with high baseline homocysteine levels from participation in LY231 514 clinical
`trials (see submission serial number 194 to IND # 40,061 dated November 24,
`1999). This letter was sent to all LY231 514 investigators except for investigators
`in two studies (Ii3E-MC-JMAF and l-13E-MC-JMAS) where patients are currently
`receiving fôlic acid supplementation. In the interest of patient safety, this action
`was taken preceeding formal protocol amendments.
`
`In the cover letter to the FDA accompanying the November 24 letter Lilly stated
`that the exclusion of patients with high baseline homocysteine levels was a
`preliminary action, Lilly also indicated that a further communication would be
`sent to the FDA with details of the updated safety analysis together with a plan
`for an intervention to lessen serious toxic effects in patients with high baseline
`homocysteine levels. The updated safety analysis (see attachment) again
`reinforces the relationship between high baseline homocysteine levels and the
`potential for serious toxicity after treatment with LY231 514 as shown by the
`following:
`
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99791
`
`TX 330
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Food and Dnig Administration
`December 3, 1999
`
`Page 2
`
`Elevated homocysteine level at baseline is highly correlated with severe
`toxicity (e.g. neutropenia, neutropenia accompanied by infection,
`diarrhea).
`
`Drug-related death is highly correlated with severe toxicity.
`
`Preliminary data from an ongoing study in gastric cancer (H3E-MC-
`JMAF) in a small number of patients has shown that at standard doses
`(500 mg/rn2), LY231 514 has demonstrated activity in the presence of folic
`acíd supplementation.
`
`LY231 514 may be escalated as high as 925 mg/rn2 when accompanied
`by folio acid supplementation at a schedule of 5 mg orally daily for two
`days before, the day of, and two days after LY231 514 administration
`(Study H3E-MC-JMAS). Data from this study shows that in patients with
`elevated homocysteine levels, this amount of supplementation causes
`homocysteine levels to drop below 10 g.iM and causes no change in
`patients that do not exhibit an elevated baseline homocysteine level.
`
`lt is well known that elevated homocysteine levels are an indicator of
`poor nutritional status, and recently, elevated homocysteine levels have
`also been shown to be a predictor of mortality in cardiovascular disease.
`A study of homocysteine levels in 1788 middle-aged and elderly.
`volunteers has shown that an elevated homocysteine level of 14 i.LM or
`greater puts a patient at increased risk for death caused by
`cardiovascular disease [Annals Internal Medicine 131:321-330, 1 991.
`
`Folic acid supplementation of 400 ig daily in elderly patients with
`elevated homocysteine levels has been shown to substantially reduce
`piasma homocysteine levels within two weeks. The level continues to
`drop slightly for another two weeks, and then plateaus [International
`Journal for Vitamin and Nutrition Research 69:187-93, 1999]. Brouwar
`and coworkers showed that low dose folic acid (250 ig - 500 .ig)
`intervention significantly decreases homocysteine levels. M eight week
`weshout period was not, sufficient for blood folate and plasma
`homocysteine levels to return to baseline. [American Journal of Clinical
`Nutrition 69:99-104, 1999] Niyikiza et al have shown that in patients who
`are not supplemented, homocysteine levels do not change over the
`course of treatment with LY231 514 [Annals of Oncology, Supplement 4,
`9: Abstract 609, 1998].
`
`Based on this analysis, we have concluded that the following interventions are
`appropriate for ongoing trials.
`
`copy
`
`CONFIDENTIAL
`ELAP00199792
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Food and Drug Administration
`December 3, 1999
`
`Page 3
`
`For the ongoing phase I trials of LY231514 in combination with another cytotoxic
`agent where the top dose level has been reached:
`
`Any patient with a baseline homocysteine level 1 2 .iM will be excluded
`from enrollment. For these trials, this is a continuatiOn of the policy
`mentioned in the November 24, 1999 letter to investigators (IND
`submission 194).
`
`In all other trials, the following actions are being implemented. The initial action
`to exclude patients with high baseline homocysteine levels was taken to prevent
`serious toxicity to a sub-set of potential patients. Further analyses and
`discussions with a number of external consultants led to the conclusion that an
`overall safety benefit to patients might be better served through the addition of
`dietary supplementation to patients receiving LY231 514 rather than exclusion of
`certain high-risk patients. Thus the following actions are being implemented.
`
`Because there is a strong link between severe toxicity and elevated
`homocysteine levels as well as severe toxicity and drug-related death, the
`strategy of reducing homocysteine levels prior to treatment with LY231 514 in
`these trials is being implemented. The following actions are being taken via
`another 9etter to investigator" to promote the immediate interest of patient
`safety. Formal protocol amendments will be submitted as soon as possible. The
`implementation schema is as follows:
`
`Each patient will have a blood sample drawn for the measurement of a
`baseline homocysteine level.
`
`Each patient Will begin a daily supplement of 350 1000 .tg folic acid with
`500 .tg being the recommended dose. In countries where folic acid is not
`available, a multMtamin containing 350 - 1000 .tg folic acid will be an
`acceptable substitute. Thìs supplementation will continue daily as long as
`the patient is on study.
`
`At this time, each patient will receive 1000 .g vitamin B12 as an
`intramuscular injection. (As mentioned in the introduction of the attached
`safety analysis, elevated homocysteine may also be caused by vitamin
`612 deficiency in é small percentage of patients.) This will be repeated
`after every nine weeks cycle as long as the patieritimains on study.
`
`After at le? severi days of folic acid suppImentation, patients may
`begin to rçeive treatment with 1X23151 '3
`
`It is recommended that any patient currently ori study begin folic acid and
`B12 supplementation as well Any patntfalIing into this category may
`receive his or her next regularly schedjatd dose of LY231514 providing
`
`r
`
`COPY
`
`CONFIDENTIAL
`ELAPOU 199793
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Food and Drug AdminsIration
`December 3, 1999
`
`Page 4
`
`he or she has had at least two days of folic acid supplementation and one
`B12 injection.
`
`For purposes of safety and study integrity, these steps Mf I be impièmented
`immediately in the ongoing randomized phase Ill study in malignant
`mesothelioma (H3E-MC-JMCH, A Single-blind Randomized Phase 3 ThaI of
`MTA plus Cisplatin versus Cisplatin in Patients with Malignànt Pleural
`Mesothelioma; see attachment for the letter being sent to the JMCH
`investigators). This will provide a unique opportunity to study the effects of folio
`acid and B12 supplementation on toxicity and efficacy in a homogerious patient
`population. The implementation steps for all other studies will be addressed
`immediately following completion of modifications in this randomized trial.
`
`Re-evaluation of the prevalence of toxicity will be performed after 100 -150
`patients receiving the proposed supplementation of folicacid and 512 have
`been enrolled and have received at least two cycles of treatment.
`
`Possible alternative methods to supplement folic acid and B12 wiN be evaluated
`and may be proposed for future thais.
`
`Therefore Lilly asks DODP for a prompt review of these proposals to supplement
`LY23l 514 patients with folio acid and vitamin B12. We believe that these actions
`will promote patient safety and not adversely effect the primary and secondary
`outcomes of the LY231514 registration trial for mesothelioma wt-iere
`approximately 40% of patients have high baseline homocysteine levels. In the
`JMCH study (the mesothelioma registration study), we are implementing the
`administration of folio acid and vitamin B12 to all patients in the LY231 514 plus
`cisplatin arm as well as the cisplatin afone arm to preserve the integrity of this
`single-blind study. At the present time almost half of the planned number of
`patients (280) are enrolled in study JMCH. At the end of the study, we intend to
`compare the toxicity and other outcomes of those patients supplemented with
`folio acid arid vitamin B12 versus those patients who did not receive this
`supplementation
`
`Because of the Implications for patient safety and also because of the
`potential effects on a registration trial Ully respectfully requests that we
`be notified by December 10 regarding DODP concurrence with these
`actions. We are available any time for a teleconference should DODP want
`to discuss these actions. Lilly is in the process of amending protocols for
`the LY231 514 studies and awaIt DODP comments before these protocol
`amendments are finalized.
`
`Lilly is closely monitoring patient safety and attempting to find potential
`predictors that may identify patients at increased risk The identification of
`
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99794
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Food and Diug Administration
`December 3, 1999
`
`Page 5
`
`elevated baseline homocysteine level as a predictor for toxicity and the steps
`taken to prevent this are a potential advance for chemotherapy with LY231 5'14.
`We thank DODP with their continuing assistance regarding the development of
`LY231 514.
`
`Please call Mr. John Worzalla at (317) 276-5052 or myself at (317) 277-3799 if
`there are any questions.
`
`Sincerely,
`
`ELI ULLY AND COMPANY
`
`Gregory T. Brophy, PhD.
`Director
`U.S. Regulatory Affairs
`
`Enclosures (2)
`LY231 514 (MTA) Safety Analysis
`Copy of Letter to lnvestigator for JMCH
`
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99795
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`LY231 514 (MTA) Safety
`Analysis
`
`03 December 1999
`
`Page 1 of 20
`
`12/03/99
`
`copy
`
`CONFIDENTIAL
`ELAPOOI 99796
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`1.
`
`Introduction
`
`1.1. Objective
`A preliminary analysis on the relationship between homocysteine lee1s and toxicity in
`patients treated with LY23 1514 (MTA), has been presented by Niyikiza et aL [1] The
`current document will provide additional information in the form of a more thorough
`analysis and will serve as the basis for discussion of a number of recommendations for
`increasing the safety of patients treated with LY23 1514. A number of retrospective,
`exploratory analyses have been performed in an attempt to identify factors which may put
`certain patients at risk of experiencing severe toxicity or death caused by treatment with
`LY231514. These include multivariate, descriptive, and contingency analyses. The
`dataset used for each analysis will be described.
`
`1.2. Background In formation
`Given the toxicity associated with the administration of anticancer drugs, the ability to
`predict those patients at risk of developing serious toxicity would represent an important
`therapeutic advance in this area. This safety analysis has been undertaken in order to
`identify potential prognostic factors which may predispose patients to experience severe
`toxicity or even death secondary to treatment with LY23 1514, a multitargeted antifolate.
`Historical data on the toxicity of lometrexol (another antifolate) and the efforts to
`modulate its toxicity, have been examined to more precisely define a starting point for the
`ana1yis of current LY23 1514 data.
`
`As might be predicted with ari antifolate, phase I studies of LY23 1514 identified
`myalosuppression as the principal dose limiting toxicity, with other toxicities including
`stornatitis, fatigue anti rash. A broad-spectrum phase 11 program with a Q3 week schedule
`of administration began in 1995 [reviewed in 2, 3,4]. The safety profile in 209 patients
`treated in this setting is consistent with the phase I experience and is reviewed in [21.
`Consistent with the toxicity profile of other antifolates, the most severe and dose-limiting
`toxicity is rnyelosuppression, while nonbematologic toxicities including mucositis,
`diarrhea, vomiting and infection can also occur. The incidence of CTC grade ifi and IV
`neutropenia is approximately 50% [2] and the rate of possibly or probably drug-related
`death to date is approximately 3.8%.
`
`Measures which affect plasma folate levels would also be expected to modulate the
`pharmacology of an antifolate. Indeed, the in vivo toxicity of lometrexol (LMTX), a pure
`purine biosynthesis inhibitor is profoundly affected by supplemental folic acid. Mice fed a
`folate-depleted diet are very sensitive to the toxicity of lometrexol whereas under folate
`replete conditions toxicity is reduced. Careful control of folate levels allows preservation
`of antitumor activity, but if sufficient excess folate is provided, antitumor activity is
`ablated [5, 6].
`
`These predlinical observations of the importance of folie acid in modulating the toxicity
`of lometrexol in animals were subsequently confirmed in phase I clinical trials. Without
`
`Pagc 2 of 20
`
`12/03/99
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99797
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`folie acid supplementation, the dose-limiting toxicity is myeiosuppression, but toxicity is
`cumulative. Consistent with the in vivo observations, supplementing patients with oral
`folie acid improves tolerance to LMTX. On a Q3 week schedule in the absence of folic
`acid, the MiD is less than 12 mg/rn2 [7]. However, the same schedule with 14 days of
`supplemental oral folie acid at 5 mg/day has a recommended phase II dose of at least 130
`mg/rn2 [8].
`
`As with lometrexol, in vivo experiments with LY23 1514 have suggested that
`supplemental folic acid modulates its toxicity profile and antitumor activity. The LD5O of
`LY23 1514 occurred at 60- and 250-fold lower doses of LY23 1514 in DBA/2 and CDI
`nu/nu mice maintained on a low folate diet compared with those fed standard diets. In
`these experiments, the antitumor activity ofLY2315J4 was preserved [91.
`
`Ongoing LY23 1514 triaIs include a phase I study of LY23 1514 and folic acid. An interim
`report suggests that folic acid supplementation in this study permits dose escalation by
`ameliorating toxicity since heavily and minimally pretreated patients tolerate LY23 1514
`at doses of 700 and 92$ mg/rn2 respectively [IO]. In a phase II trial in gastric cancer, a
`small set of patients has also received folie acid supplementation. This trial will be
`discussed in further detail in Section 3.
`
`Given the relevance of folic acid to the pharmacology of antifolates, it is reasonable to
`postulate that functional folate status could be an important predictor of toxicity. Recent
`studies have suggested that plasma homocysteine is a much more sensitive measure of
`functional folate status than is serum folate or red blood cell folate. [Ii, 12] Methionine
`synthase is a highly folate dependent enzyme that converts homocysteine to methionine.
`Thus, under conditions of folate deprivation, levels of plasma homocysteine increase.
`Elevated serum homocysteine can also result from cobalarnin (vitamin B12) or vitamin
`B6 deficiency. This is because cobalamin is a required cofactor for methionine synthese
`and the enzyme cystathionine synthase, in converting homocysteine to cystathionine, uses
`vitamin B6 as a cofactor, Serum methylmalonic acid levels are elevated in cobalamin
`deficiency, but not in folate deficiency and therefore it is important to measure levels of
`methylmalonic acid to distinguish between cobalantin and folate deficiency.
`Cystarhionine levels are markedly elevated in vitamin B5 deficiency and are elevated to a
`lesser extent in both cobalarnin and folate deficiency.
`
`As the findings which linked folate status to toxicity with lonietrexol were coming to
`light, the development of LY23 1514 was well into the decision phase, je, early phase II
`studies were taking place in a variety of solid tumors for the purposes of evaluating
`antitumor activity. In response to this information, the protocols for these ongoing studies
`were amended to collect vitamin metabolite levels at baseline and at various timepoints
`throughout the study. As such, within the entire LY23 1514 patient database, only a subset
`of patients have data on vitamin metabolite levels.
`
`Page3of2O
`
`12103199
`
`COPY
`
`CONFIDENTIAL
`ELAPOO 199798
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Initial Multivariate Analysis
`1.2.1.
`In late 1997, an analysis was conducted to assess the relationship of vitamin metabolites,
`drug exposure, and other pre-specified baseline patient characteristics to toxicity
`following therapy with LY23 1514. [11 Data was examined from 139 phase II patients
`with tumors of the colon, breast, pancreas, and esophagus who had been treated with
`LY23 1514 at 600 mg/rn2 1V over 10 minutes once evexy 21 days. These patients had
`homocysteine (Hcys), cystathionine, and methylmalonic acid levels measured at baseline
`and once each cycle thereafter. Stepwise regression modeling, multivariate analysis of
`variance, and discriminant analysis were implemented.to determine which predictors
`might correlate with severe toxicity, and to predict which patients are at high risk of
`experiencing such toxicity. Prognostic factors considered were age, gender, prior therapy,
`baseline albumin, liver enzymes, ANC, platelets, vitamin metabolites, and AUC.
`
`Statistically significant predictors of Grade 4 neutropenia (n=21 pts) were albumin
`(p<zO.0001) and Ilcys (p=O.002O). Grade 4 thrombocytopenia (n=8) was predicted by
`Hcys (p<OE0001) and Albumin (p=0.0237). Hcys was also found to be the only significant
`(p=0.00l4) predictor of Grade 3/4 mucositis, diarrhea, rash, or fatigue following one
`course of therapy with LY23 1514. Hcys lOpM predicted Grade 4 neutropenia in cycle
`one 75% of the time. Hcys alone in 71% of cases predicted grade 4 neutropenia Hcys and
`albumin levels did not appear to change from baseline during therapy with LY231514.
`While AUC was not found to be a predictor of toxicity, little variability was observed in
`AUC. Maximum values were still below AUC values related to hematological toxicity in
`phase I studies. These findings led to the following conclusions:
`
`Toxicity resulting from therapy with LY23 1514 appears to be higher in patients with
`elevated pretherapy homocysteine levels.
`Elevated baseline homocysteine levels (10j.tM, for the 139 patients included in this
`analysis) highly correlate with severe hematological and nonhematologic toxicity
`following therapy with LY23 1514.
`Homocysteine was found to be better than albumin at predicting toxicity and was not
`altered with LY23 1514 therapy.
`
`1.3. Overview of LY231514 Database
`Approximately 1127 patients have been treated with LY23 1514 to date. This number
`includes:
`
`880 patients treated in "in-house" studies with full safety data available
`
`Multivariate analyses on predictors for drug-related death have been performed
`using data from these 880 patients. (34 drug related deaths in this group.)
`
`Of these 880 patients, a subgroup of 305 patients have data on homocysteine
`levels Multivariate analyses on predictors for toxicity have been performed using
`data from these 305 patients.
`
`Page 4 of 20
`
`12/03199
`
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99799
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`In addition to the analyses using multivariate methods, a number of contingency
`analyses (je, 2 x 2 tables) have been undertaken for the purposes of corroborating the
`multivariate findings.
`
`247 patients treated in "external" studies with limited data available. (9 drug-related
`deaths in this group.)
`c> Although no multivatiate analyses can be run over these patients, they will be
`included in descriptive analyses.
`
`Patient distribution by tumor type is shown in Table 1.1.
`
`Table 1.1
`
`PatIent Disposition: Distribution of Patients by Tumor Type
`Tumor Type
`
`Number of Patients
`
`Phase] Single Agent
`
`Bladder
`Breast
`
`Cervix
`
`CRC
`H&N
`JMAS'/JMAW
`
`NSCLC, front line
`
`NSCLC, second line
`
`Other 01
`Pancreas
`Ph 1 Combination
`
`Renal
`
`Total*
`
`100
`
`29
`
`123
`
`35
`
`203
`
`35
`
`40/24
`
`160
`
`81
`
`29
`
`42
`
`187
`
`39
`
`1127
`
`*Note: Mesothelioma trials are not included in this number.
`Folic acid supplementation study
`Rena] impairment study
`
`Because the trials in mesothelioma have recently started and the data are immature, they
`have not been included in the overall safety analysis.
`
`2.
`
`Updated Multivariate Analyses
`
`21. Analysis for Drug-related Death
`A total of 43 deaths in 1127 patients have been reported as possibly or probably related to
`LY23 1514, giving an overall death rate of 3.8%. In describing any on-study death,
`
`Page5of20
`
`12/03/99
`
`COPY
`
`CONFIDENTIAL
`ELAPOO 199800
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`investigators select a designation of probably, possibly, or not related to treatment with
`LY231514. The 43 deaths discussed represent the most conservative approach to
`designation of causality, and include all deaths that have been designated as possibly or
`probably related to treatment. An internal review has indicated that four of these (patients
`403-4089,403-4102, 403-4103, and 720-7003) may be attributed to other causes such as
`disease progression and/or underlying medical conditions. When these patients are
`excluded from the total, the rate of drug-related death is 3.5%. Nevertheless, we have
`performed all analyses discussed here using all deaths designated by the investigator as
`possibly or probably drug-related. The prevalence of drug-related death by tumor type is
`shown in Table 2.1.
`
`Table 2.1. Prevalence of Drug Related Death by Tumor Type
`
`Tumor Type
`
`Bladder
`
`Breast
`
`Cervix
`
`CRC
`
`Head and Neck
`NSCLC 1' Line
`NSCLC 2" Line
`JMAS (FA Supplement)
`
`JMAW (Renally impaired)
`Other Gr
`Pancreas
`Phase i Sngle Agent
`Phase i Combination
`
`Renal
`
`Overall Prevalencè
`'EsophagaI, gastric
`
`Number of Drug Related Deaths (% of
`Total Pts)
`2 (6.9%)
`2(1.6%)
`1 (2.9%)
`3(1.5%)
`4(11.4%)
`2(1.2%)
`8(9.9%)
`0(0%)
`
`1 (4.2%)
`
`6(20.1%)
`0 (0%)
`6(6%)
`4(2.1%)
`4(10.3%)
`
`43(3.8%)
`
`Total
`
`29
`
`123
`
`35
`203
`
`35
`
`160
`
`81
`
`40
`
`24
`
`29
`
`42
`
`100
`
`187
`
`39
`
`1127
`
`This table demonstrates the observation that particular tumor types may have a higher
`death rate. Although the sample sizes are small, these tumors appear to be bladder, head
`and neck, esophageal, gastric, renal, and second-line non-small cell lung cancer. Patients
`with these tumors might be expected to be nutritionally or metabolically deficient,
`lending support to the theory that nutritional status may play a large role in the degree of
`toxicity or death attributable to LY23 1514. In a second cluster of tumors which includes
`breast, colorectal, front-line non-small cell lung, and pancreas cancers, the rate of death is
`approximately 1.6%.
`
`Page6of2o
`
`12/O3/9
`
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99801
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`2.1,1. Definition of Patient Populatton
`Of the patients treated with LY23 1514, 880 of these have been treated on "in-house"
`studies, and have complete data available. Within this dataset, there have been 34 drug-
`related deaths.
`
`The remaining patients (247) have been treated in studies performed by contract research
`organizations. Data from these studies have not been integrated into the main database,
`and therefore, these patients are not included in any multivariate analyses. For any
`descriptive statistics, the group of 1127 patients will be used as the denominator.
`
`2.1.2. Methods
`Stepwise regression modeling, multivariate analysis of variance, and discriminant
`analysis were implemented to determine which predictors might correlate with drug-
`related death. Prognostic factors considered for LY23 15 14-related deaths were age,
`gender, baseline albumin, liver enzymes, ANC, platelets, AUC, pEe-treatment weight,
`prior treatment, tumor type, grade 4 neutropenia in conjunction with grade 3 or 4
`infection (a surrogate indicator of febrile neutropenia), post baseline minimum platelet
`count, grade 3 or 4 diarrhea, and grade 3 or 4 mucositis.
`
`Note** Because of the small number of patients with data(n =11 of 34), vitamin
`metabolítes (including homocysteine) were not included in the multivariate analysis
`for LY231514-related deaths.
`
`2.1.3. Results
`Results of these analyses are shown in Table 2.2. Those factors that are independent
`predictors for drug-related death ata statistically significant level are shown in bold.
`
`Page7of2ø
`
`12103199
`
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99802
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Table 2.2. Prognostic Factors for LY231514 Related Deaths (n
`
`880)
`
`Variables
`
`Age
`
`Gender
`
`BLALB
`
`BL ALT
`
`BL ALK Phos
`
`BL PLT
`
`BL ANC
`AUC*
`
`WEIGHT
`
`Prior Treatment
`
`G3!4 IN + G4Neut
`Tumor Type
`0314 MUC
`G314 Diarrhea
`
`P value
`
`0.1009
`
`0.6286
`
`0.1182
`
`0.4591
`
`0.9884
`
`0.6106
`
`0,3385
`
`0.0759
`
`0.9893
`
`0.4169
`
`<0.00001
`
`0.0019
`0.5164
`
`0.0016
`
`*c1culared from population PK model [13))
`Abbreviations: BL ALB, baseline albumin; BL ALT, baseline atanine transaminase; BL ALK
`Phos, baseline alkaline phosphatase; BL PLT, baseline platelets; BL ANC, baseline absolute
`neutrophil count; AUC, area under the curve; IN, infection; Neut, neutropenia; MUC, mucositis.
`
`As expected, Grade 3/4 infection accompanied Grade 4 neutropenia, Grade 4 neutropenia,
`Grade 3/4 diarrhea and tumor type were all significantly associated with drug-related
`death.
`
`In cant ngency analyses designed co compare the rate of death in patients receiving a
`starting dose of 500 mg/rn2 vçrsus 600 mg/rn2, it was found that there was no difference
`in the relationship between death and either starting dose.
`
`2.1.4. Homocysteine Leve?s in LY231514-refated Death
`Homocysteine levels are currently available in only 11 patients of the 43 whose deaths
`were reported as possibly or probably related to MTA. Table 2.3 shows the baseline
`homocysteine levels in these patients.
`
`Page 8 of 20
`
`12/03/99
`COPY
`
`CONFIDENTIAL
`ELAPO01 99803
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Table 2.3.
`
`Baseline Homocysteine Levels ¡n 11 Patients Who Experienced
`Possibly or Probably Drug-related Death
`
`Patient Number (tumor)
`Baseline Homocysteine
`1-1 (CRC)
`22.5
`4-155 (CRC)
`13.3
`505-29 (H&N)
`14.4
`802-19 (H&N)
`12.4
`802-26 (H&N)
`31.9
`803-849 (breast)
`14,3
`100-5002 (renal impairment)
`5.2
`4034l02* (2 line NSCLC)
`10
`403.4103* (2' line NSCLC)
`3.7
`801.8006 (ph I wfdocetaxel)
`3.6
`302-3035 (ph I w/oxalipiatin)
`12.8
`An internal review has indicated that these deaths may have been due to disease progression and/or
`underlying medical conditions.
`
`2.2. Analyses for Drug-related Toxicity
`Because the initial multivariate analysis suggested that honiocysteine levels were
`critically important in predicting toxicity, these updated analyses for toxicity have been
`performed on the set of 305 patients who have had their baseline homocysteine levels
`measured and recorded. All homocysteine levels were measured using a single laboratory.
`
`Table 2.4 shows the incidence of selected hematologic and nonhematologic toxicities
`according to tumor type.
`
`L
`
`L
`
`Page 9 of 20
`
`12103/99
`COPY
`
`CONFIDENTIAL
`ELAPOOI 99804
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Table 2.4. Prevalence of Selected Ilemalologic and Nonhematologic Toxicìtles by Tumor Type
`
`Tumor Type
`
`G3/4 Diarrhea
`(%)
`
`G4 Neul
`(%)
`
`G314 INF
`(%)
`
`Bladder
`Breast
`
`Cervix
`
`CRC
`
`NSCLC V'
`
`NSCLC 2
`
`3 (10.3%)
`
`4(3.2%)
`
`2(5.7%)
`
`9(5.26%)
`0(0.0%)
`
`3(3.1%)
`
`7 (24%)
`
`35 (28.5%)
`
`11(31.4%)
`
`31(18.1%)
`14(40%)
`
`16(16.7%)
`
`I (3.5%)
`5(4.1%)
`
`0(0.0%)
`
`4(2.3%)
`6(17.1%)
`1(1%)
`
`G4 PLT
`(%)
`
`0 (0.8%)
`
`12(9.8%)
`2(5.7%)
`
`11(6.4%)
`
`4(11.4%)
`
`3 (3.1%)
`
`G3/4 INF +
`G4Neut
`(%)
`1 (3.5%)
`
`3(2.4%)
`0(0.0%)
`
`2(1.2%)
`
`3(8.6%)
`
`0(0.0%)
`
`1(1.2%)
`
`0 (0.0%)
`
`G314 MUC
`(%)
`
`G2/3/4 MUC
`(%)
`
`0 (0.0%)
`
`9 (7.3%)
`
`4(11.4%)
`
`2(1.2%)
`
`5(14,3%)
`
`4 (4.2%)
`
`g (27.6%)
`
`29(23.6%)
`
`10(28.6%)
`
`15(8.8%)
`
`12(34.3%)
`
`19(19.8%)
`
`3 (3.7%)
`
`JMAS
`
`JMAW
`
`Other 61*
`
`Phi Ccmbo
`
`Renal
`
`Total
`
`0(0.0%)
`0(0.0%)
`
`0(0.0%)
`
`5 (7.0%)
`
`8 (5.9%)
`
`3 (7.7%)
`
`15 (18.5%)
`5(12.5%)
`
`3(12.5%)
`
`24 (33.8%)
`
`58 (42.5%)
`
`11(28.2%)
`
`4 (5%)
`0(0.0%)
`
`0(0.0%)
`4 (5.6%)
`
`6 (4.4%)
`
`3 (7.7%)
`
`6(7.4%)
`2(5%)
`
`1 (4.2%)
`
`12 (16.9%)
`
`4 (2.9%)
`
`6(15.4%)
`
`0(0.0%)
`0(0.0%)
`
`1(4.2%)
`
`6 (8.4%)
`
`3 (2.2%)
`
`1(2.6%)
`
`0(0.0%)
`
`4 (5.6%)
`
`5 (3.7%)
`
`3 (7.7%)
`
`3 (7.5%)
`3(12.5%)
`
`IO (14.1%)
`
`13 (9.6%)
`
`5 (12.8%)
`
`37(4.2%)
`
`230(26%)
`
`34(3.9%)
`
`22(2.5%)
`
`63(1.2%)
`
`35(4.0%)
`
`130 (14.8 %)
`
`'1'Pancreas, esophageal, gastric
`
`r)
`C
`
`r,C)
`
`oQ-n1(DZ
`Q>(Ji'-
`
`Page 10 of 20
`
`12103199
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`2.2.1. Definition of pntient population
`Of the patients with solid tumors treated with LY23 1514, 305 had homocysteine (Hcys)
`levels measured at baseline. A subgroup of these also had cystathionine and
`methylmalonic acid measured at baseline and once each cycle thereafter. To eliminate
`the complicating factor of the effect of folic acid supplementation on toxicity, any patient
`who received folic acid supplementation (n =38 in Study JMAS) was removed from the
`analysis, leaving a final sample size of 267.
`
`2.2.2. Methods
`Stepwise regression modeling, multivariate analysis of variance, and discriminant
`analysis were implemented to determine which predictors might correlate with severe
`toxicity, and to predict which patients are at high risk of experiencing such toxicity.
`Prognostic factors considered were age, gender, baseline albumin, liver enzymes, ANC,
`platelets, vitamin metabolites, pre-treatment weight, and AUC, tumor type, and prior
`treatment. The B12 metabolite methyl malonic acid is highly correlated with
`homocysteine and was therefore removed from the analysis to eliminate issues of
`colinearity,
`
`2.2.3. Results
`Results of these analyses are shown in Table 2.5. Those factors which are significant
`predictors for toxicity are shown in bold.
`
`Page 11 of 20
`
`12103199
`
`copy
`
`CONFIDENTIAL
`ELAPOOI 99806
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Table 2.5.
`
`Variables
`
`PrognostIc Factors for Kematologic Toxicity
`(n = 267)
`
`G4 Neutropenia
`+ G314 Infection
`
`G4 Neutropenia Throinbocytopenia
`
`Grade 314
`Diarrhea
`
`0.5120
`0.8705
`
`0.5391
`
`0.7246
`
`0.0452
`
`0.6136
`
`0.1932
`
`0.3423
`
`0.2206
`
`0.9044
`
`Age
`
`Gender
`
`BL ALB
`
`BL ALT
`
`BL ALK Plias
`
`BL IICYS
`
`BL CYST
`BLPLT
`
`BL ANC
`
`AUC
`Weight
`
`Prior Treatment
`Tumor Type
`
`0.9735
`
`0.2528
`
`0.6348
`
`0.6916
`
`0.3874
`
`<0.00001
`
`0.8030
`
`0.5250
`
`0.6029
`
`0.7298
`
`0.6487
`05059
`0.4855
`
`0.8050
`0.5208
`
`0.1934
`
`0.6050
`
`0.0573
`
`0.0191
`
`0.5971
`
`0.8101
`
`0. 2737
`
`0.6081
`
`0.3182
`
`0.8122
`0.0153
`
`<0.00001
`
`<0.00001
`
`0.3907
`
`0.4457
`
`0.0736
`
`0.9531
`
`0.0633
`
`0.4813
`0.1315
`
`0.9454
`
`0.2066
`
`0.2345
`
`0.3204
`
`0.99 18
`
`0.478 8
`0,4305
`
`Results of this analysis show that baseline bomocysteine level is a statistically significant
`predictor for febrile neutropenia, Grade 4 neutropenia, Grade 4 thrombocytopenia, and
`Grade 3 or 4 diarrhea.
`
`These results confirm the findings of the original multivariate analysis (discussed in the
`introduction), and support the conclusion that homocysteine may provide an ideal
`prognostic variable for predicting toxicity during LY23 1514 therapy.
`
`2.3.4. Contingency Malysis for LY231514-related Toxicity
`In order to investigate the relationship between toxicity prevalence and baseline
`homocysceine levels, contingency tables have been prepared. Table 2.6 shows the
`relationship between baseline homocysteine and the percent of patients experiencing
`toxicity.
`
`Page 120(20
`
`12/03/99
`
`copy
`
`CONFIDENTIAL
`ELAP0OI 99807
`
`Lilly Ex. 2103
`Sandoz v. Lilly IPR2016-00318
`
`

`
`Table 2.6. Prevalence of Toxicity by Baseline Homocysteine Threshold
`
`Toxicity
`
`n
`
`n
`
`Baseline Honweysteine Threshold
`Threshold 10 pM Threshold = Il pM Threshold
`12 pM Threshold = 13 uM Threshold =14 11M
`i2
`13
`10
`11
`l4
`<12
`<13
`<14
`<10
`<11
`n = 80
`n = 51
`n = 242
`u =63
`n = 187
`n = 254
`n = 36
`n 269
`225
`118
`43%
`Any hem' tox + feb neut
`48%
`24%
`22%
`52%
`57%
`24%
`25%
`61%
`26%
`Any hem + nonhenit Lox
`48%
`24%
`64%
`27%
`53%
`61%
`27%
`59%
`28%
`29%
`41%
`45%
`30%
`G 314 ANC in Cycle 1
`40%
`38%
`32%
`50%
`33%
`33%
`32%
`40%
`G 4 ANC on shdy
`44%
`48%
`24%
`21%
`53%
`56%
`24%
`23%
`25%
`48%
`G 3/4 ANC on study
`46%
`58%
`63%
`58%
`47%
`72%
`48%
`48%
`69%
`0314 Jnf on study
`8%
`6%
`0%
`1%
`8%
`1%
`1%
`1%
`6%
`11%
`04 ANC and G 3/4 mf on study
`0%
`05%
`5%
`8%
`0.5%
`6%
`1%
`8%
`1%
`11%
`G 4 AI9C and 03/4 Diarrhea on study
`6%
`2%
`2%
`6%
`2%
`2%
`8%
`1%
`10%
`8%
`4%
`4%
`4%
`4%
`22%
`4%
`G4ANCandG2/3/4Mucositisonstudy
`25%
`12%
`17%
`14