Am J Hum Genet 37:1-31, 1985
`
`The Natural History of Homocystinura Due to Cystathionine
`jP-Synthase Deficiency
`S. HARVEY MUDD,' FLEMMING SKOVBY,1 HARVEY L. LEVY,2
`KAREN D. PETTIGREW,3 BRIDGET WILCKEN,4 REED E. PYERITZ,5 G. ANDRIA,6
`GODFRIED H. J. BOERS,7 IRVIN L. BROMBERG,8 ROBERTO CERONE,9
`BRIAN FOWLER,'0 H. GROBE,11 HILDGUND SCHMIDT,12 AND LESLIE SCHWEITZER13
`
`SUMMARY
`An international questionnaire survey has been conducted to define better
`the natural history of homocystinuria due to cystathionine P-synthase
`deficiency and permit evaluation of treatment. Data were compiled for
`629 patients. Among patients not discovered by newborn screening, B6-
`responsive individuals on the average have significantly better mental
`capabilities (mean IQ, 79) than do B6-nonresponsive individuals (mean
`IQ, 57). Time-to-event curves are presented for the other major clinical
`abnormalities produced by this disease. Each occurred at significantly
`lower rates in untreated B6-responsive than in untreated B6-nonresponsive
`patients, as shown by the following examples: (1) dislocation of optic
`lenses (at age 10, chances of dislocation: 55% and 82%, respectively);
`(2) initial clinically detected thromboembolic events (at age 15, chances
`
`Received May 29, 1984; revised August 10, 1984.
`1 Laboratory of General and Comparative Biochemistry, National Institute of Mental Health, Bldg.
`32, Rm. 101, 9000 Rockville Pike, Bethesda, MD 20205.
`2 The Joseph P. Kennedy, Jr. Laboratories of the Neurology Service, Massachusetts General Hospital,
`and the Department of Neurology, Harvard Medical School, Boston, MA 02114.
`3 Theoretical Statistics and Mathematics Branch, National Institute of Mental Health, Bethesda,
`MD 20205.
`4N.S.W. Department of Health, Oliver Latham Laboratory, North Ryde 2113, Sydney, Australia.
`5Departments of Medicine and Pediatrics, The Johns Hopkins University School of Medicine,
`Baltimore, MD 21205.
`6 Clinica Pediatrica della II Facolta di Medicina e Chirurgia, Napoli, Italy.
`7Department of Medicine, University of Nijmegen, Nijmegen, Netherlands.
`8 The Hospital for Sick Children, Toronto, Ontario M5G 1X8, Canada.
`9 Clinica Pediatrica R ((G. Gaslini)), Universita di Genova, Genova, Italy.
`10 Willink Biochemical Genetics Unit, Royal Manchester Children's Hospital, Manchester M27
`IHA, England.
`Kinderklinik der Universitat Mtinster, D-4400 Munster, West Germany.
`12 Kinderklinik der Universitat Heidelberg, D-6900 Heidelberg, West Germany.
`13 New York State Institute for Basic Research in Developmental Disabilities, Staten Island, NY
`10314.
`© 1985 by the American Society of Human Genetics. All rights reserved. 0002-9297/85/3701-0001$02.00
`1
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`2
`
`MUDD ET AL.
`of having had such an event: 12% and 27%, respectively); (3) radiologic
`detection of spinal osteoporosis (at age 15, chances of such osteoporosis
`having been detected: 36% and 64%, respectively); and (4) mortality
`(at age 30, chances of not surviving: 4% and 23%, respectively). Me-
`thionine restriction initiated neonatally prevented mental retardation,
`retarded the rate of lens dislocation, and may have reduced the incidence
`of seizures. Pyridoxine treatment of late-detected B6-responsive patients
`retarded the rate of occurrence of initial thromboembolic events. Fol-
`lowing 586 surgical procedures, 25 postoperative thromboembolic com-
`plications occurred, six of which were fatal. Reproductive histories
`were reported predominately for B6-responsive patients. Living offspring
`of either men or women patients had few abnormalities. The evidence
`is inconclusive whether untreated maternal cystathionine P-synthase
`deficiency leads to excessive fetal loss. Only 13% of patients detected
`in screening programs of newborns and classified as to B6-responsiveness
`were B6-responsive, compared to 47% among late-detected patients.
`Current screening programs that identify neonatal hypermethioninemia
`may be preferentially failing to detect B6-responsive patients.
`
`INTRODUCTION
`Homocystinuria due to cystathionine P-synthase deficiency is a genetically de-
`termined inborn error of the transsulfuration pathway biochemically characterized
`by increased plasma homocyst(e)ine and methionine and decreased cyst(e)ine.
`The disease was discovered in 1962, when mentally retarded individuals were
`screened for abnormal urinary amino acids [1, 2]. Two years later, the enzyme
`defect, deficient activity of cystathionine P-synthase, was demonstrated [3]. *
`The major clinical manifestations include mental retardation, dislocation of the
`optic lens (ectopia lentis), skeletal abnormalities, and a tendency to thromboembolic
`episodes [6].
`Once the enzyme defect and the major biochemical aberrations had been defined,
`dietary therapy based upon methionine restriction and L-cystine supplementation
`was suggested. This regimen resulted in some degree of control of the biochemical
`abnormalities, and largely anecdotal evidence emerged that such treatment from
`the newborn period could prevent or delay clinical manifestations [7-10]. Sub-
`sequently, it was found that some patients on normal diets respond biochemically
`to large doses of vitamin B6t with decreases in plasma homocyst(e)ine and changes
`of plasma methionine and cyst(e)ine concentrations toward normal [11], while
`others do not so respond. Evidence now available strongly indicates that such
`
`* Several other enzyme lesions are now known that also lead to excretion of excess homocystine
`(homocystinuria) [4, 5]. In this paper, we will deal only with homocystinuria due to cystathionine
`,B-synthase deficiency, and the condition will be designated by the latter name.
`t Vitamin B6 will hereafter be referred to as B6 or pyridoxine.
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`HOMOCYSTINURIA
`
`3
`
`B6-responsiveness, or lack thereof, is one manifestation of a considerable hetero-
`geneity in the mutations producing deficiencies of cystathionine P-synthase activity
`[6]. Since 1967, many patients have been given trials or prolonged periods of
`dietary and/or pyridoxine therapy. These therapeutic trials were carried out in
`the absence of any randomly selected, untreated control population. More recently,
`the use of aspirin and dipyridamol has been suggested to prevent thrombosis
`[12], but neither has been subjected to rigorous testing. Other workers have
`advocated the administration of betaine to reduce concentrations of homocysteine
`[13, 14].
`Only a few years elapsed between discovery of cystathionine P-synthase de-
`ficiency and initiation of various therapies. Consequently, there was little op-
`portunity to accumulate knowledge about the natural history of the condition. It
`is clear, however, that the age of onset and the severity of clinical manifestations
`vary widely among affected individuals. Thus, the prevalence and natural history
`of each of the pleiotropic features remain uncertain. Few data are available upon
`the impact of maternal cystathionine P-synthase deficiency on reproductive potential
`and on the fetus. The effects of genetic heterogeneity, as indicated, for example,
`by B6-responsiveness or B6-nonresponsiveness, on each of these manifestations
`are largely undefined. These gaps in our knowledge impede realistic assessment
`of the efficacy of various therapies. Since homocystinuria due to cystathionine
`P-synthase deficiency is a relatively rare disease [6], no single physician or center
`has accumulated a sufficiently large sample of patients to address these questions.
`Therefore, we have conducted a worldwide questionnaire survey and collected
`information in a standardized format on more than 600 patients with homocystinuria
`due to proven or presumed deficiency of cystathionine
`3-synthase. Our results,
`presented in this report, clarify some of the major uncertainties about the natural
`history of the disease and the role of genetic heterogeneity. They also establish
`baselines for future evaluation of the effects of treatment in this disease.
`
`METHODS
`
`Data Base
`A standardized questionnaire was designed and mailed to each clinician known from a
`previous study [15] to be caring for patients with cystathionine 0-synthase deficiency.
`Physicians were asked to complete a questionnaire for each such individual about whom
`they had appropriate information. To encourage participation, the questionnaire was kept
`relatively simple.* Each patient was identified by first name, first two letters of family
`name, birth date, and sex. Affected relatives were specified. Further questions focused
`upon the factor(s) that led to ascertainment, whether the patient was responsive to B6,
`and upon the presence and age of appearance of major clinical manifestations. A detailed
`history of therapy was requested, as well as a reproductive history. To permit use of rel-
`evant published material, the respondent was asked to identify articles concerning a given
`patient. Additional sources of information were identified by a review of the literature
`and by contacting centers around the world specializing in diagnosis and management of
`inborn errors of metabolism. Further, physician cooperation was solicited by notices in
`
`* Copies of the questionnaire are available through the National Auxiliary Publications Service
`(see footnote t to page 10).
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`4
`
`MUDD ET AL.
`
`appropriate journals. For some patients upon whom recent information could not be obtained,
`the study coordinators completed questionnaires chiefly or solely on the basis of published
`material. Such patients were included only when sufficient details were available to prove
`that they did not overlap with any patient otherwise included in the study. Data collection
`occurred during 1982 and early 1983. Data from the completed questionnaires were entered
`into a computer and verified by proofreading a print-out of the computer data against the
`original questionnaires. A computer search for duplication due to a single patient having
`been reported upon by two different physicians detected several such instances, and the
`redundant information was deleted.
`
`Statistical Analyses
`
`Time-to-event curves were calculated according to the product-limit estimate method
`of Kaplan and Meier [16]. Comparisons among the curves for statistically significant
`differences were performed by the procedures of Gehan [ 17] and Breslow [ 18]. For evaluation
`of treatments, numbers of expected events were calculated according to the nonparametric
`procedures described by Turnbull et al. [19]. Differences between numbers of observed
`and expected events were tested for statistical significance according to the same procedure.
`
`RESULTS
`The Study Population and Criteria for Acceptance into Study
`For the present survey, updated information was received concerning 532
`homocystinuric patients with proven or presumed cystathionine P-synthase de-
`ficiency. To this group was added material on an additional 97 patients obtained
`primarily from published reports [7, 9, 20-55], bringing the total to 629 patients.
`All patients admitted to the study had been demonstrated to be excreting homo-
`cystine. To restrict the population to cystathionine P-synthase-deficient individuals
`(thus excluding other causes of homocystine excretion [4-6]), either (1) cysta-
`thionine P3-synthase deficiency had to have been demonstrated directly by enzyme
`assay or (2) the patient had to have either hypermethioninemia or dislocated optic
`lenses. Table 1 shows the similar percent distributions of these findings in patients
`with and without confirmation of the diagnosis by enzyme assay.
`
`TABLE 1
`CRITERIA FOR ACCEPTANCE OF PATIENTS INTO STUDY
`
`ENZYME ASSAY
`Not performed
`Performed*
`
`............ 147 (68.0%)
`Dislocated lens and hypermethioninemia ......
`21 (9.7%)
`Dislocated lens only .............
`.......................
`Hypermethioninemia only ..........
`39 (18.1%)
`.....................
`9 (4.2%)
`Neither of above ................
`.......................
`216 (100%)
`Total ...............................................
`
`274 (66.3%)
`75 (18.2%)
`64 (15.5%)
`0 (0%)
`413 (100%)
`
`* The following tissues were used for enzyme assays (followed by no. patients, in parentheses): liver (27);
`both liver and cultured fibroblasts (13); cultured fibroblasts (170); brain (1); phytohemagglutinin-stimulated
`lymphocytes (3); transformed lymphocytes (1); cultured fibroblasts and phytohemagglutinin-stimulated lym-
`phocytes (1).
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`HOMOCYSTINURIA
`
`5
`
`TABLE 2
`CLINICAL FEATURES LEADING TO INVESTIGATION FOR HOMOCYSTINURIA
`
`Clinical feature
`
`Ectopia lentis
`.......................
`Mental retardation .4.0
`Developmental retardation .1.5
`Early thromboembolic disorder .1.1
`.0.9
`Marfanoid characteristics
`Bony abnormality
`.0.2
`Seizures .0.2
`Behavioral or psychiatric disorder
`Othert .0.4
`
`Sole cause
`(% of patients)
`
`20.6
`
`0
`
`Contributory
`cause*
`(% of patients)
`
`Total in
`which a cause
`(% of patients)
`
`65.0
`51.7
`21.0
`15.0
`36.0
`23.3
`3.0
`2.8
`10.6
`
`85.6
`55.7
`22.5
`16.1
`36.9
`23.5
`3.2
`2.8
`11.0
`
`NOTE: Based on data for 472 patients not ascertained as a result of screening of newborns or screening of all
`sibs of a proband.
`* Includes all patients with the specified feature, as well as at least one other, reported as leading to investigation
`for homocystinuria.
`t Includes a variety of manifestations, none of which was a cause in as many as 2% of the population.
`
`Of the 629 patients, 307 were females and 321 males, close to the expected
`ratio of 1:1. The sex of one patient was not specified. Sixty-four patients (10.2%)
`were dead at the time of reporting.
`
`Ascertainment
`Data on factors leading to investigation of patients for homocystinuria were
`available for 618 patients. Of these, 58 were discovered during screening of
`newborns, and an additional 88 were discovered by screening all siblings after
`detection of homocystinuria in a proband, leaving 472 patients ascertained on
`the basis of clinical features. Table 2 displays the frequencies at which each of
`the major clinical manifestations was the sole reported cause of investigation for
`homocystinuria or was a contributory cause. Most patients were investigated
`because of more than one clinical feature, the major exception being the almost
`21% initially investigated solely because of ectopia lentis.
`
`B6-Responsiveness
`Of the 629 patients, 231 (36.7%) were classified as biochemically responsive
`to B6 when not folate depleted; 231 (36.7%) were classified as nonresponsive to
`B6; 67 (10.7%) were judged intermediate in response; and 100 (15.9%) had not
`been classified. For subsequent analyses in this presentation, neither the "inter-
`mediate-response" group, although this may include patients with a biochemically
`significant response, nor unclassified patients were included in groups designated
`as "B6-responsive" or "B6-nonresponsive."
`The relative frequencies of B6-responsive and B6-nonresponsive patients among
`at least two subgroups of the total population differed markedly from the overall
`frequency. Among the 55 patients who had been both discovered by newborn
`screening and classified with respect to B6-responsiveness, seven (12.7%) were
`B6-responsive, 43 (78.2%) were nonresponsive, and five (9.1%) were intermediate
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MUDD ET AL.
`6
`in response. Thus, the ratio of responders to nonresponders was 1:6 in this subgroup.
`Among 25 patients who were not hypermethioninemic when untreated and who
`had been classified with respect to B6-responsiveness, 21 (84%) were responsive,
`two (8%) were nonresponsive, and two (8%) were intermediate in response,
`yielding a ratio of responders to nonresponders of 10:1.
`For many sibships, more than one affected member had been classified as to
`B6-responsiveness. Table 3 shows that among such sibs there was almost complete
`concordance of responsiveness or nonresponsiveness. In no case was a patient
`judged fully responsive when one of his or her sibs was judged fully nonresponsive.
`
`Hypermethioninemia
`Among the patients for whom data were available concerning the presence or
`absence of hypermethioninemia in the untreated state, 524 (93.6%) were hyper-
`methioninemic and 36 (6.4%) were not. As expected from the relative prepon-
`derance of B6-responsive patients among those who were not hypermethioninemic,
`10% of untreated B6-responders were not hypermethioninemic, whereas less than
`1% of untreated B6-nonresponders were not hypermethioninemic.
`
`Mental Capabilities
`Responding physicians were asked to rate the mental capabilities of patients
`in several ways:
`(1) IQ. A plot of the distribution of IQ's among all patients for whom such
`data were available is shown in figure 1. To eliminate any effect of very early
`therapy, data on patients discovered by newborn screening were not used in
`constructing these plots of IQ or in the alternative analyses of mental capabilities
`discussed in the following sections. Figure 1 demonstrates a very wide range in
`patient IQ's-from 10 to 138. The median of the cumulative frequency curve
`was at an IQ of approximately 64. The curve for patients classified as B6-responders
`was shifted toward higher IQ's (median 78), whereas that for patients classified
`as B6-nonresponders was shifted toward lower IQ's (median 56). As a result,
`only about 4% of B6-nonresponders had IQ's of 90 or above, but 22% of B6-
`responders had values in this range. The difference between the mean IQ for B6-
`
`TABLE 3
`CONCORDANCE OF B6--RESPONSIVENESS IN SIBLINGS
`
`ADDITIONAL SIBS:
`B6-RESPONSIVENESS
`
`FIRST SIB: B6-RESPONSIVENESS
`Int. *
`Yes
`No
`
`56
`
`Yes .
`Int.* ..1
`No ...................
`M.
`57
`
`...
`10
`...
`10
`
`...
`
`2
`50
`52
`
`y
`
`56
`13
`50
`119
`
`NOTE: Based upon 104 sibships (total of 223 sibs) in which more than one
`sib had been classified with respect to B6-responsiveness.
`* Int. = intermediate.
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`HOMOCYSTINURIA
`
`7
`
`o80
`
`20
`
`80
`80
`20
`40
`CUMULATIVE FREQUENCY
`OF PATIENTS, percent
`FIG. 1.-Distributions of IQ's among patients not detected by newborn screening. The curves are
`based on the following nos. of patients with reported IQ's: all B6 categories, 284; B6-responders,
`107; B6-nonresponders, 115. For any specified IQ value, the plots show the total percent of each
`category of patients with IQ's equal to, or less than, the specified value.
`
`100
`
`responders (79) and that for nonresponders (57) was highly significant (P S
`.0001).
`(2) If no IQ value was available, physicians rated the patients as being either
`"grossly retarded," "mildly retarded," 'average or above average intelligence,
`but with a learning disability," or "average or above average intelligence." Table
`4 shows the proportions of patients in each of these categories. The trends are
`clearly the same as those resulting from IQ measurements, with 61 % of the
`patients being either grossly or mildly retarded, but only 38% of B6-responders
`being so rated, and almost 85% of B6-nonresponders showing retardation. The
`difference between the distributions of B6-responders and nonresponders was
`again highly significant (P -
`.0001).
`(3) If a proband had one or more documented nonhomocystinuric sibs, responding
`physicians were asked to rate the proband's intelligence as either "below," the
`"same" as, or "above" that of unaffected sibs. The results, table 4, follow a
`.0001) higher ratings for B6-
`similar trend, again with very significantly (P -
`responders than for B6-nonresponders.
`To evaluate further the validity of the results described above, for each rating
`method, the analyses were repeated after prior division of the patients into subgroups
`consisting of those with a diagnosis confirmed by enzyme assay and those without
`such confirmation. No substantial differences were noted between these two
`subgroups.
`To gain some insight into the influence of ascertainment bias on the ratings of
`mental capabilities, the above analyses were repeated (1) after removal of the
`few patients ascertained solely on the basis of mental retardation and (2) after
`removal of all patients with mental retardation included as one factor in ascer-
`tainment. The first procedure produced virtually no change; the second led, as
`expected, to shifts toward higher IQ's among all groups. The medians of the
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`8
`
`~~~~~MUDDET AL.
`
`>
`0
`
`o.
`
`Cu~~~~~~~~~~~~~~~~~1
`4)~~~
`~~Cu
`.0
`
`W LL.
`0<
`
`Cu3
`2
`
`C1
`
`F
`
`u 0
`'z
`
`0
`
`.0
`
`04) cn W
`
`C>
`Q~
`.0
`
`c
`0 c
`
`U
`
`00
`~~~~~~~~~~0
`
`w 0~~~
`
`r- e1-
`
`<
`
`F>
`
`Cu3
`
`C's
`
`00
`
`r
`
`40
`
`~~~~~~~~~CO00
`00
`
`4
`
`00
`
`0~~~~~~~0C
`
`C40
`
`Cu
`
`0
`
`4)Cu0
`Cu~~0Cu ~~Cu
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`HOMOCYSTINURIA
`
`9
`
`A
`A
`
`8
`
`A
`
`0
`
`0
`
`00
`A
`AA
`AA
`80
`A
`A
`A A 0AO
`A A
`1A
`A0A
`a 0-A AA4
`A
`A 80 AA A
`A4LA -A&
`A A A
`
`A
`
`20
`
`A
`
`0
`
`0
`
`Pan
`
`aA
`A
`
`A
`A
`Panel A
`
`0
`
`A
`
`100
`
`8
`
`O
`
`so8
`
`20 -A
`
`80
`
`4
`Ao
`A
`
`A
`
`A
`
`24
`
`000 0
`0
`A
`A
`AA
`0
`68A A A
`-A
`0
`
`A
`
`A
`hA
`
`A~AA
`
`A
`
`A
`
`A
`
`0
`
`AA
`
`A A
`A
`A yA
`A A
`
`Panel B
`
`10
`
`12
`
`8L
`
`AGE, years
`FIG. 2.-Effect of early treatment on IQ of B6-nonresponsive patients. B6-nonresponsive patients
`ascertained by newborn screening and treated with methionine restriction from early ages are each
`represented by an open circle. The same patients are plotted in both panels A and B. The ages are
`those at which the IQ's were measured. The comparison populations consist of B6-nonresponsive
`patients not ascertained by newborn screening. Each such patient is represented by a closed triangle.
`For comparison populations, the ages are: panel A, age at ascertainment; panel B, age at last follow-
`up. The lines are regression lines calculated for the IQ's of the comparison populations: panel A,
`IQ = 57.27 - (0.01567) x (age at ascertainment, in mos); mean IQ = 55.7 ± 2.6 (SEM); panel
`B, IQ = 58.81 - (0.01827) x (age at last follow-up, in mos); mean IQ = 57.6 ± 1.9 (SEM).
`
`cumulative IQ curves rose to 81 for patients in all B6 categories combined, to 86
`for B6-responders, and to 64 for B6-nonresponders. Thus, the difference in median
`IQ between B6-responders and nonresponders was essentially unchanged (about
`22 points), and the difference in mean IQ remained highly significant (P S
`.0001). Similar shifts occurred in the more qualitative alternative estimates of
`mental capabilities, and again the differences between B6-responders and non-
`responders remained highly significant (data not shown).
`General experience with cys-
`Effect of early treatment on mental capability.
`tathionine P-synthase-deficient patients has shown that late treatment rarely, if
`ever, completely reverses mental impairment [6], although methionine restriction
`or treatment with pyridoxine [56, 57] or betaine [13] have been reported to lead
`to behavioral improvement and moderate increases in IQ [57], suggesting a re-
`versible component to the mental disturbance of the untreated disease [57]. Early
`treatment, however, might prevent the mental damage [10]. The effect of early
`treatment is illustrated in figure 2. IQ's for B6-nonresponsive patients identified
`as neonates and treated from very early ages with methionine restriction, usually
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`10
`
`MUDD ET AL.
`accompanied by L-cystine supplementation (early-treated), are compared to IQ's
`of B6-nonresponsive patients not detected by newborn screening (late-detected).
`If treated at all, very few of the latter patients had commenced therapy at ages
`of less than 1-2 years. IQ's of early-treated patients are plotted at the age of
`measurement of the latest reported IQ. The ages of measurement of the IQ's for
`the late-detected patients were not specified in the questionnaire. These IQ's are
`plotted alternatively as a function of the age at ascertainment (panel A) or the
`age at last follow-up (panel B). It is apparent that for these late-detected patients
`severe impairment of IQ was manifest from early ages, and this impairment did
`not change markedly as a function of age. * The IQ's of the early-treated patients
`are higher, with little overlap with the late-detected groups at ages up to 7 to 8
`(mean IQ = 94 ± 4 [SEM]; P < .001 compared to mean IQ of either late-
`detected group).
`Qualitative estimates of mental capabilities for the early-treated group are in
`agreement with the above results. Of 23 patients without IQ measurements, one
`(4%) was estimated to be grossly retarded, three (13%) mildly retarded, one (4%)
`average with learning disability, and 18 (78%) to be average or above. These
`values are very favorable compared to those in table 4 for late-detected B6-
`nonresponsive patients.
`For the 15 patients detected by newborn screening who were B6-responsive,
`or were intermediate or unclassified as to B6 response, reported IQ's ranged from
`82-110 (six patients), and all other patients were estimated to be of "average or
`above average" intelligence. Most of these patients had been treated with me-
`thionine restriction from the time of diagnosis, several had received B6 in addition,
`and four (three responders, one intermediate) had been treated solely with vitamin
`B6. Again, these results are highly favorable compared to those in table 4.
`Dislocation of Optic Lenses
`Data on ages of lens dislocation were analyzed according to the Kaplan and
`Meier method, with the important stipulation that for each patient the only interval
`considered was that prior to initiation of treatment specific for cystathionine 1-
`synthase deficiency. Hence, a patient was removed from the group at risk for
`lens dislocation (censored) at the time he or she started such treatment. The result-
`ing time-to-event graphs are shown in figure 3.t These plots demonstrate that,
`
`* The apparent tendency of IQ in the early-detected group to decrease after age 8 may to some
`extent be due to a change in the method used to determine IQ. Prior to age 6, methods that measure
`only verbal ability (e.g., Stanford-Binet) are generally used; thereafter, methods that evaluate per-
`formance as well as verbal ability (e.g., WISC) are used increasingly. Some patients we have observed
`received a markedly higher "verbal" than "performance" score on the WISC test and, correspondingly,
`had lower overall IQ scores at ages when this test was used than at a younger age. The questionnaire
`did not ask the responding physician to specify the type ofIQ test used, and further data would be
`needed to evaluate fully the effects of this possible methodological change.
`t The numerical values upon which this figure is based have been deposited with, and are available
`through, the National Auxiliary Publications Service. See NAPS document no. 04244 for 17 pages
`of supplementary material. Order from NAPSc/o Microfiche Publications, P.O. Box 3513, Grand
`Central Station, New York, NY 10163. Remit in advance in U.S. funds only $7.75 for photocopies
`or $4.00 for microfiche. Outside the U.S. and Canada, add postage of $4.50. $1.50 for microfiche
`postage.
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`HOMOCYSTINURIA
`
`11
`
`a
`
`F
`4 4
`U)
`
`50
`
`-
`
`< y ills ark
`
`B6 - non-respons*ve
`o~~~~~~~ANtyrm combined
`
`0 30-
`
`z U
`
`Ui20-
`U).1zLU
`
`14-
`
`10
`
`7-
`
`5-
`
`3
`
`0c
`
`c
`0L
`
`zL
`
`U
`cc
`LU0.
`
`4
`
`8
`
`12
`
`16
`
`24
`
`28
`
`32
`
`36
`
`40
`
`20
`AGE, Years
`FIG. 3.-Time-to-event graphs for lens dislocation in untreated patients. Patients were removed
`from the at-risk groups upon commencement of any therapy. Probabilities were calculated according
`to the Kaplan and Meier method [16] and are plotted on a logarithmic scale. Time points are plotted
`only when they are the first or last for a given probability. Based on data for following nos. of
`patients: B6-responsive, 231; B6-nonresponsive 231; all types, 628. Each plot is discontinued at the
`time the no. of patients remaining at risk decreases to 10. For clarity, the graph for "all patients"
`is plotted starting at age 4.
`
`for all groups of patients, there was a lag period of approximately 2 years before
`appreciable lens dislocation occurred. After age 2, lens dislocations began to be
`detected, but at different rates in B6-nonresponders and B6-responders, so that
`50% of untreated nonresponsive patients had dislocated lenses by age 6, whereas
`for untreated B6-responsive patients, 50% dislocation was attained at approximately
`age 10. After age 3, smoothed logarithmic plots are reasonable approximations
`of straight lines for both groups. Thus, during this phase, approximately half of
`the B6-nonresponsive patients who start an interval with intact lenses can be
`expected to dislocate during the next 40 months, while for untreated B6-responders,
`half can be expected to dislocate each 95-100 months. Statistically, the difference
`between the time-to-event graphs for B6-responders and B6-nonresponders was
`highly significant (P < .0001). Note, however, that these plots most correctly
`represent the detection of lens dislocation rather than its actual occurrence. If a
`patient had dislocated lenses at ascertainment, and his or her history did not
`establish a prior time of dislocation, then dislocation was placed at the time of
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`12
`
`MUDD ET AL.
`ascertainment. As a result, these curves represent maximum ages. Serial exam-
`inations of untreated patients from early ages might well result in curves shifted
`to earlier ages in such plots.
`Effect of treatment on lens dislocation. An important property of the Kaplan-
`Meier curves shown in figure 3 is that they furnish a statistical prognosis for an
`untreated patient starting at any specified age. This permits assessment of the
`effect of treatment on lens dislocation in an assorted group of patients who start
`treatment at different ages and have been maintained on treatment for different
`intervals. For each such patient, one can calculate the probability that, during
`the interval he or she is at risk on treatment, lens dislocation would have occurred
`had he or she not been treated.* The combined probabilities for a group of such
`patients yield a predicted number of dislocations if they had been untreated or if
`treatment is without effect. This number can be compared to the actual number
`of dislocations that occurred during treatment. Table 5 displays such comparisons.
`Among 24 B6-responsive patients with lenses intact who were given pyridoxine
`(with or without folate) as initial and continuous treatment (column A), 8.4
`dislocations would have been expected to occur during treatment, as calculated
`from the time-to-event graph (fig. 3) if treatment had no effect. Five dislocations
`occurred. This decrease is not statistically significant (P > .05). Column C takes
`into account additional experience by including intervals of treatment that took
`place after temporary interruption of the initial treatment or in which B6 therapy
`followed earlier periods of dietary treatment. The decrease in the number of lens
`dislocations observed below those expected now became statistically significant.
`In table 5, B6-nonresponsive patients are separated into "late-detected" and
`"early-treated" (i.e., those identified by newborn screening). No effect of therapy
`was observed among the few "late-detected" B6-nonresponsive patients who had
`intact lenses at the time they were started on dietary therapy. Among "early-
`treated"patients, however, for each treatment regimen evaluated, the number of
`dislocations observed was significantly below the number expected (P c
`.001),
`suggesting that therapy may be beneficial when begun in early infancy. Since six
`of these patients had suffered lens dislocations, the benefit may be more in delaying
`the time of lens dislocation rather than in absolute prevention. Follow-up of these
`patients to older ages is necessary to clarify this issue.
`
`Thromboembolic Events
`From the thromboembolic events originally reported, nine described as "pos-
`sible" or "partial" and two that were only questionably related to the underlying
`condition were deleted. There remained a total of 253 events, occurring in 158
`patients. No reported events occurred in 471 patients. Of these 253 events, 81
`(32%) were cerebrovascular accidents, 130 (51 %) affected peripheral veins (with
`
`* The interval at risk is the time from onset of treatment in a patient with lenses not yet dislocated
`until the earliest of one of three events: (1) treatment termination, (2) time of last follow-up (if the
`patient is on continuing treatment), or (3) lens dislocation. The probability that an untreated patient
`will develop dislocation during the interval tl to t2 is (1 -P2/Pi), where pi and P2 are the probabilities
`that lenses are not dislocated (fig. 3) at times tj and t2, respectively.
`
`Lilly Ex. 2096
`Sandoz v. Lilly IPR2016-00318
`
`

`
`HOMOCYSTINURIA
`
`13
`
`C
`
`>
`
`0
`
`C
`
`cq
`
`r-
`
`i.>
`
`4)
`>
`?
`0
`
`C
`
`C
`
`.0
`
`C
`
`=
`
`> C,3
`
`>0
`
`O'o
`
`C
`0
`
`C0
`
`O 2
`
`C0
`
`C
`
`0 0
`a :3 0
`r
`lo.,z
`C
`0
`Q 0 0
`C
`
`0
`
`C > IC S.
`
`- m
`
`Cd
`
`C
`V
`0 -,=