Special Article
`
`0
`
`0
`
`m
`
`©
`
`©
`
`©
`
`Prevalence of cobalamin deficiency in the Framingham
`elderly populationl-a
`
`John Lindenbaum, Irwin H Rosenberg, Peter WF Wilson, Sally P Stabler, and Robert H Allen
`
`ABSTRACT To determine whether the increased prevalence
`of low serum cobalamin concentrations in elderly people repre-
`sents true deficiency, serum concentrations of cobalamin and fo-
`late and of metabolites that are sensitive indicators of cobalamin
`deficiency were measured in 548 surviving members of the orig-
`inal Framingham Study cohort. Serum cobalamin concentrations
`< 258 pmol/L were found in 222 subjects (40.5%) compared
`with 17.9% of younger control subjects (P < 0.001). Serum
`methylmaionic acid and total homocysteine concentrations were
`markedly elevated in association with cobalamin values < 258
`pmol/L in 11.3% and 5.7%, respectively, of the cohort. Both
`metabolites were increased in 3.8% of the cohort, associated with
`significantly lower erythrocyte counts and higher mean cell vol-
`umes. Serum metabolites correlated best with serum cobalamin
`values, even when subnormal determinations were excluded. The
`prevalence of cobalamin deficiency was -> 12% in a large sample
`of free-living elderly Americans. Many elderly people with "nor-
`mal" serum vitamin concentrations are metabolically deficient
`in cobalamin or folate. Am J Clin Nutr 1994;60:2-11.
`
`KEY WORDS Cobalamin deficiency, folate deficiency,
`methylmalonic acid, homocysteine, vitamin B-12, Framingham
`Study, elderly people
`
`Introduction
`
`ciency (23, 24), and elevated serum Hcys concentrations are in-
`creasingly recognized as a major risk factor for cerebral, coro-
`nary, and peripheral vascular disease (25-27). Therefore it is
`important to determine whether true deficiency of Cbl is a sig-
`nificant problem in elderly people, especially because it is easily
`treatable and its effects are reversible if detected in time.
`Serum concentrations of methylmalonic acid (MMA) and total
`Hcys have proven to be highly sensitive indicators of tissue de-
`ficiency of cobalamin (21, 23, 28-32). The concentration of one
`or both metabolites was found to be markedly elevated (> 3 SD
`above the mean in normal control subjects) in 99.8% of a con-
`secutive series of > 400 patients with clinically proven Cbl de-
`ficiency (32). Furthermore, > 5% of patients with clear-cut he-
`matologic or neurologic disorders (or both) caused by lack of the
`vitamin have serum Cbl values in the lower end of the normal
`range, even though circulating metabolite concentrations are un-
`equivocally elevated (31). Increases in serum metabolites are also
`frequently seen before the development of low Cbl values in
`patients with pernicious anemia in remission, when maintenance
`vitamin B-12 injections are not given for several months (31).
`Although MMA accumulates in Cbl but not folate deficiency,
`Hcys concentrations are also increased in folate deficiency (23,
`30, 32, 33), and the combined measurement of both metabolites
`facilitates the differentiation of these disorders (32, 34).
`
`Over more than three decades, many investigators have re-
`ported a strikingly increased prevalence of low serum cobalamin
`(Cbl; vitamin B-12) concentrations in elderly people (1-16), al-
`though a minority have disputed this finding (11, 17-19). Some
`have regarded the decreased Cbl values in geriatric populations
`as a laboratory phenomenon of little or no clinical significance
`or as the result of bias in patient selection. The majority of elderly
`individuals with low serum concentrations have not been anemic
`nor have they developed megaloblastic anemia after follow-up
`(4, 15). On the other hand, Cbl deficiency may cause serious
`neuropsychiatric damage, including impairment in cognitive
`function, in the absence of anemia (20-22). Furthermore, ho-
`mocysteine (Hcys) and its derivatives accumulate in Cbl deft-
`
`~ From the Department of Medicine, Columbia-Presbyterian Medical
`Center and the College of Physicians and Surgeons, Columbia Univer-
`sity, New York; the United States Department of Agriculture Human
`Nutrition Research Center on Aging at Tufts University and Tufts Uni-
`versity School of Medicine, Boston; the Framingham Heart Study, Fra-
`mingham, MA; and the Departments of Medicine and Biochemistry, Bio-
`physics and Genetics, University of Colorado Health Sciences Center,
`Denver.
`2 Supported by Department of Health and Human Services research
`grants DK-21365 and AG-09834; the National Institute of Diabetes and
`Digestive and Kidney Diseases; and the National Institute on Aging.
`3 Address reprint requests to John Lindenbanm, Department of Med-
`icine, Columbia-Presbyterian Medical Center, New York, NY 10032.
`Received September 30, 1993.
`Accepted for publication January 5, 1994.
`
`Am J Clin Nutr 1994;60:2-11. Printed in USA. © 1994 American Society for Clinical Nutrition
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1105-0001
`
`

`
`COBALAMIN DEFICIENCY IN ELDERLY PEOPLE
`
`¯ Frarningham Subjects, Age 65-99 (N=548)
`
`[] Control Subjects, Age 22-63 (N=117)
`
`30
`
`20
`
`10
`
`0- 74- 148- 221- 295- 369- 443- 516- 590- 664- Z738 (pmo~L)
`0- 100- 200- 300- 400- 500- 600- 700- 800- 900- zl000 (pg/ml)
`Semm Cobalamin
`
`FIG 1. Distribution of serum cobalamin concentrations in 548 elderly subjects from the Framingham Heart Study
`and 117 younger control subjects.
`
`We therefore decided to assess the problem of whether there
`is a significant increased prevalence of cobalamin deficiency in
`elderly people by measuring both vitamins and both metabolites
`in a large sample of a well-defined ambulatory population--the
`surviving members of the original Framingham Heart Study.
`
`Subjects and methods
`
`The study was approved by the Human Investigations Review
`Committee at the New England Medical Center.
`The Framingham study, an epidemiologic investigation of
`heart disease, was established in Framingham, MA, between
`1948 and 1950 (35). The original cohort consisted of 5209 sub-
`jects 29-63 y of age. The surviving members of this cohort have
`been examined every 2 y. In 1988-89 ~--1200 survivors partici-
`pated in the 20th examination. Serum was available for assay of
`vitamin and metabolite concentrations from 548 of the first 567
`people to report for that examination. Healthy young medical
`personnel 22-63 y of age employed by Columbia University
`served as control subjects for the serum vitamin assays. There
`were 117 control subjects (59 females, 58 males; median age 30
`y) for the Cbi assay and 59 (30 females, 29 males; median age
`30 y) for the folate assay. Fifty healthy blood donors (25 males,
`25 females) from Denver, ranging in age from 18 to 65 y (median
`35 y), served as control subjects for the metabolite assays, which
`were performed in Denver (30).
`Blood was obtained from the Framingham subjects at 1300 h
`and maintained in the refrigerated state for 18-24 h, followed
`by centrifugation for 30 rain at 2500 x g at 4 *C and separation
`of serum, which was stored at -20 *C. Blood specimens were
`obtained from the younger control subjects between 0900 and
`1300 h and centrifuged for 30 rain at 2500 × g at 4 *C after l h
`
`at room temperature; the serum was stored at -20 °C. Andersson
`et al (36) reported modest and essentially equivalent increments
`in plasma Hcys concentrations when whole blood was centri-
`fuged after 24 h of refrigeration or after 1 h at room temperature.
`Separate aliquots of frozen sera were thawed once and assayed
`for vitamin or metabolite concentrations 2-21 mo after collec-
`tion. Metabolite concentrations are stable in serum maintained at
`-20 °C for many years (23, 28, 32).
`Serum Cbl and folate concentrations were determined by si-
`multaneous radioassays using purified intrinsic factor and milk
`folate binder, respectively (Quantaphase; Bio-Rad Laboratories,
`Richmond, CA). Serum MMA and total Hcys were measured by
`modifications (30, 37) of techniques using capillary gas chro-
`matography and mass spectrometry (38, 39). An automated com-
`plete blood count including platelet count and mean cell volume
`(MCV) was performed as well as measurements of serum elec-
`trolytes, blood urea nitrogen (BUN), creatinine, transaminases,
`alkaline phosphatase, lactic dehydrogenase, albumin, total pro-
`tein, calcium, phosphate, and magnesium. Creatinine clearance
`was calculated by a formula including serum creatinine, age, and
`body weight (40). Serum antibodies to intrinsic factor (4 i ) were
`measured when additional serum was available from subjects
`with laboratory tests suggestive of Cbi deficiency.
`Information regarding vitamin supplementation was available
`from a questionnaire (42) mailed to all of the subjects that was
`returned by 401 members (73.2%) of the cohort. The number of
`years of formal education and the stated average amount of daily
`alcohol consumption was available for the entire cohort.
`Bivariate associations between variables were tested by cal-
`culating Pearson linear-regression coefficients of correlation.
`Log-transformed data were used for serum MMA, Hcys, Cbl,
`folate, creatinine, and BUN because the distribution of each of
`these variables was skewed toward higher values. For multivari-
`
`Th :I. s One
`
`FDGO-CgJ-ssT1
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1105-0002
`
`

`
`LINDENBAUM ET AL
`
`ate analysis of factors predicting serum MMA and Hcys, a step-
`wise multiple linear-regression model (BMDP2R system) was
`used (43) to select the most strongly related predictor variables
`and to calculate a multiple unadjusted R2, with an F of 4 for entry
`or removal.
`
`Results
`
`Elderly population
`
`The 548 subjects included 348 women and 200 men aged 67-
`96 y (£ ___ 1 SD 77.4 _+ 6.2, median 77.5 y). All were white.
`Results of automated biochemical screening tests were normal in
`most subjects. Elevations of serum transaminases were found in
`21 of the 548, which were modest (< 200 U/L) in all but two
`instances. The serum creatinine was increased (> 124 #mol/L in
`men, > 106/zmol/L in women) in 47 (8.5%) of the 548 subjects,
`> 177 /.tmol/L (2.0 mg/dL) in 9, and > 265 /zmol/L in 3. The
`calculated creatinine clearance was > 80 rnL/min (a value often
`used clinically as a lower limit of normal in younger patients) in
`only 15.4% of the cohort and was < 60 mL/min in 51.3%.
`
`Cobalamin deficiency
`
`The mean (_+SD) serum Cbl concentration in the 548 members
`of the Framingham elderly cohort (315 ___ 145 pmol/L, 427 ___ 196
`pg/mL) was lower than that of the 117 healthy control subjects
`<65 y of age (375 ___ 124 pmol/L, 508 ± 168 pg/mL; P
`< 0.001 ). In the elderly group, 29 subjects (5.3%) had Cbl values
`< 148 pmol/L (200 pg/mL), the manufacturer’s stated lower
`limit of normal for the radioassay used, in contrast with the 1.7%
`for the control subjects (X2 = 2.8, P > 0.05). Serum concentra-
`tions < 258 pmoi/L (< 350 pg/mL) were found in 222 (40.5%)
`of the elderly subjects compared with 21 (17.9%) of the control
`subjects (X2 = 21.2, P < 0.001). The distribution of Cbl concen-
`
`trations was generally shifted toward lower values in the elderly
`cohort (Fig 1).
`The serum MMA concentration was increased (> 376 nmol/
`L, or 3 SD above the control mean) in 62 of the 222 elderly
`subjects with serum Cbl concentrations < 258 pmol/L (Fig 2).
`These 62 subjects constituted I 1.3% of the entire cohort of 548
`individuals. Unless otherwise indicated, the terms increased and
`elevated will be used throughout this article to denote values for
`MMA and/or Hcys > 3 SD above the mean in normal control
`subjects 18-65 y of age. In the clinical assessment of patients
`with megaloblastic anemias, we have found that values 3 SD
`above the mean provide greater specificity in the diagnosis of the
`vitamin deficiency states, despite the sacrifice of some sen-
`sitivity (32).
`The serum total Hcys value was elevated (> 21.3/~mol/L, 3
`SD above the control mean) in 31 of the 222 elderly subjects
`with Cbl concentrations < 258 pmol/L (5.7% of the entire co-
`hort; Fig 2). In 72 subjects with a Cbl value < 258 pmol/L
`(13.1% of the cohort), either the MMA and/or Hcys concentra-
`tion was increased and in 21 (3.8%) both values were elevated.
`The two highest values for both MMA and Hcys were in the
`two persons in whom serum Cbl was < 74 pmol/L (< 100 pg/
`mL) (Fig 2). In subjects with elevated metabolite concentrations,
`the degree of elevation was similar in those with serum Cbl val-
`ues between 74 and 147 pmol/L as in those with values in the
`range 148-258 pmol/L (Fig 2).
`
`Folate deficiency
`
`The mean serum folate concentration in the 548 elderly sub-
`jects did not differ from that of the control subjects < 65 y of
`age [22.0 nmol/L (9.7 ± 8.3 ng/mL) vs 17.4 nmol/L (7.7 ± 3.4
`ng/mL), respectively; P > 0.05]. In 16 (2.9%) of the 548 elderly
`subjects serum folate was < 5.9 nmol/L (2.6 ng/mL), the man-
`ufacturer’s stated lower limit of normal for the radioassay, in
`
`10,000
`
`5,000
`
`2,000
`
`I ,ooo
`
`5o0
`
`200
`
`100
`
`50
`
`20
`
`80 .
`
`7O
`
`........ ~- ....
`
`~-~ +3SD
`
`-2SD
`
`lO
`
`~ ÷2SD
`
`-2SD
`
`3~7 74 1’11 148 1~5
`
`5b 1~0 1;0 2;0 2~0
`Serum Cobalamin
`
`2’~1
`
`3~0
`
`2~,8 (pmoI/L)
`
`3~0 (~ml)
`
`0
`
`~
`
`3~7
`
`~0
`
`7=4
`
`1~0
`
`1~1 I;8 1~5 2~’1 2~8 (Drool/L)
`
`1~0 2~0 2~0
`Serum Cobalamin
`
`3~0 a~o (~/ml)
`
`FIG 2. Serum methylmalonic acid and total homocysteine concentrations in elderly Framingham subjects with serum
`cobalamin concentrations < 258 pmol/L (< 350 pg/mL).
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1105-0003
`
`

`
`COBALAMIN DEFICIENCY IN ELDERLY PEOPLE
`
`contrast to 1 (1.7%) of the 59 control subjects, (P > 0.05). Serum
`concentrations < 11.3 nmol/L (< 5.0 ng/mL) were found in 129
`(23.5%) of the elderly and 14 (23.7%) of the control subjects
`(P > 0.05).
`Serum Hcys was elevated in 19 of the 129 elderly subjects
`with a serum folate concentration < 11.3 nmol/L (3.5% of the
`entire cohort). However, in 17 of these 19 subjects, serum Cbl
`was also < 258 pmol/L, accompanied in 10 instances by an el-
`evation of serum MMA.
`There was a tendency for low serum concentrations of the two
`vitamins to be associated with each other. Serum folate was
`< 11.3 nmol/L in 82 (36.9%) of the 222 subjects with a Cbl
`concentration < 258 pmol/L, in contrast to 47 (14.4%) of 326
`subjects with a Cbl concentration > 258 pmol/L (X2 = 37.2, P
`< 0.001). Serum Cbl was < 258 pmol/L in 82 (63.6%) of the
`129 subjects with a serum folate < 11.3 nmol/L. In the entire
`cohort, serum Cbl correlated with serum folate (r = 0.48 for the
`log of each serum concentration, P < 0.001). Neither the folate
`nor Cbl concentration correlated with the hematocrit, MCV,
`stated amount of alcohol intake, or number of years of education.
`
`Serum methylmalonic acid
`
`Serum MMA was elevated in 82 (15.0%) of the 548 elderly
`subjects. Members of the cohort with increased MMA values
`were older and had significantly lower Cbl and folate concentra-
`tions and higher serum creatinine values (Table 1). In Figure 3
`the percent of subjects at various intervals of serum Cbl concen-
`trations who had MMA elevations is shown, as compared with
`the percentage of the entire cohort with an increased MMA con-
`centration (15%), as indicated by the interrupted line. Serum
`MMA concentrations were increased in > 15% of subjects with
`serum Cbl values < 258 pmol/L and in < 10% of those with
`values -> 258 pmoFL. Serum Cbl was < 258 pmol/L in 62
`(75.6%) of the 82 subjects with MMA elevations; in 10 of these
`62, serum creatinine was above normal.
`Serum Cbl was -> 258 pmol/L in 20 subjects with an elevated
`MMA. Serum creatinine was increased in 10 of them and the
`creatinine clearance was < 60 mL/min in 7 of the 10 in whom
`serum creatinine was normal. In the remaining three subjects,
`serum Cbl varied from 273 to 365 pmol/L and the creatinine
`clearance from 64 to 75 mldmin.
`The most striking MMA elevations were noted in the subjects
`with serum Cbl concentrations < 258 pmol/L. In 23 of the 62
`persons with Cbl values below this concentration and with a high
`
`metabolite concentration, serum MMA was in the range 662-
`6820 nmol/L, whereas the highest MMA value in the 20 subjects
`with serum Cbl --> 258 pmol/L was 638 nmol/L (X2 = 8.6, P
`< 0.005).
`
`Serum total homocysteine
`
`Serum Hcys was elevated in 39 (7.1%) of the 548 members of
`the elderly cohort. Subjects with increased Hcys values were
`older and had significantly lower serum Cbl and folate and higher
`serum creatinine concentrations than those without increased val-
`ues (Table 1). These differences were also significant when male
`and female members of the cohort were analyzed separately, ex-
`cept that the mean creatinine values in males did not differ (data
`not shown).
`Serum Cbl was < 258 pmol/L in 31 (79.5%) of the 39 subjects
`with Hcys elevations and was associated with an increased MMA
`concentration in 21. Nineteen (48.7%) of the 39 subjects with
`elevated Hcys values had serum folate concentrations < 11.3
`nmol/L; however, 17 of the 19 also had Cbl concentrations < 258
`pmol/L.
`Sixteen (41.0%) of the 39 subjects with elevated Hcys values
`had increased serum creatinine concentrations. Six of the eight
`subjects with Hcys elevations who had Cbl concentrations > 258
`pmol/L had increases in serum creatinine and folate values
`~ 11.3 nmoi/L. In the remaining two subjects the folate concen-
`tration was < 11.3 nmol/L and creatinine was normal.
`The combination of a serum folate concentration < 11.3 nmol/
`L and a normal MMA was seen in 9 of the 39 persons with Hcys
`elevations ( 1.6% of the entire cohort). In seven of the nine serum
`Cbl was < 258 pmol/L and in two of the seven serum creatinine
`was elevated as well. Each of the 39 subjects with an increased
`Hcys concentration had a low or low-normal Cbl or folate con-
`centration, an increased creatinine concentration, or some com-
`bination of these findings.
`
`Elevations of both metabolites
`
`Serum concentrations of both MMA and Hcys were elevated
`in 25 (4.6%) of the 548 elderly subjects. Serum Cbl was < 258
`pmol/L in 21 of the 25; 7 of the 21 also had increased serum
`creatinine. Serum creatinine was also high in each of the four
`individuals with elevations of both metabolites and a serum Cbl
`value -> 258 pmol/L. Serum folate was < 11.3 nmol/L in 10 of
`the 25 subjects in whom both MMA and Hcys were increased;
`serum Cbl was < 258 pmol/L in each of them.
`
`TABLE 1
`Serum cobalamin, folate, and creatinine concentrations in elderly subjects with and without markedly elevated metabolite concentrations/
`
`Age(y)
`Serum cobalamin (pmol/L)
`Serum folate (nmol/L)
`Serum creatinine (/zmol/L)
`
`Serum MMA
`
`Serum Hcys
`
`>3 SD
`(n = 82)
`
`80___7z
`217 _ 832
`18.1 ± 12.5J
`106 ± 532
`
`<3 SD
`(n = 456)
`
`77±6
`332 ± 146
`22.7 __. 19.5
`88 __. 27
`
`>3 SD
`(n = 39)
`
`81 _82
`197 ___ 772
`12.7 ± 8.22
`115 ± 352
`
`<3 SD
`(n = 509)
`
`77_+6
`325 __. 145
`22.9 ± 19.0
`88 _ 27
`
`i.~ ___ SD. MMA, methylamonic acid; Hcys, total homocysteine; >3 SD, 3 SDs above the mean in normal control subjects aged 18-65 y. To
`
`convert cobalamin values to pg/mL, multiply by 1.355. To convert folate values to ng/mL, multiply by 0.44. To convert creatinine values to mg/dL,
`multiply by 0.011.
`z.J Significantly different from <3 SD for same metabolite: 2 p < 0.001, ~ P < 0.05.
`
`0
`
`0
`
`m
`
`©
`
`©
`
`©
`
`o
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1105-0004
`
`

`
`5O
`
`--~ 40
`Z
`
`>- 30
`
`LINDENBAUM ET AL
`
`Percent of all 548
`subjects with elevated
`Methylmalonic acid
`
`0
`
`0
`
`m
`
`©
`
`©
`
`©
`
`148- 185-
`<148
`<200 200- 250-
`
`221-
`300-
`
`258- 295° 332- 369° 443-
`350- 400- 450- 500- 600-
`
`516- (prn01/L)
`700-(pg/ml)
`
`Serum Cobalamin
`
`n
`
`29
`
`41
`
`65
`
`87
`
`64
`
`71
`
`44
`
`62
`
`38
`
`47
`
`FIG 3. Percent of subjects with elevated serum methylmalonic acid concentrations (> 3 SD above the mean in
`younger normal control subjects) in 548 members of the Framingham elderly population with various concentrations
`of serum cobalamin, n = number of subjects.
`
`Serum vitamin and metabolite abnormalities
`
`The findings in the 74 members of the cohort with elevated
`concentrations of one or both metabolites associated with de-
`creased values for one or both vitamins are summarized in Table
`2, In 44 subjects (8.0% of the entire cohort) represented on the
`first two lines of Table 2, serum Cbi was < 258 pmoUL and the
`folate concentration was normal, accompanied by increases in
`one or both metabolite values. These subjects were likely to have
`isolated deficiency of Cbl. In 21 additional subjects (line 3, Table
`2), lack of Cbl was also highly probable, but an associated folate
`deficiency could not be excluded; in seven subjects there may
`have been a deficiency of one or the other, or of both. Serum
`creatinine was elevated in a minority of subjects in each of these
`categories. In only two subjects (0.4% of the cohort) were the
`findings clearly indicative of isolated folate deficiency.
`Because we defined an elevated serum metabolite concentra-
`tion as a value > 3 SD above the mean in a group of younger
`
`blood donors, we considered the possibility that most of the el-
`evations were not due to Cbl deficiency but merely indicated an
`upward shift in the normal range of concentrations for MMA and
`Hcys in elderly people. We therefore examined a subset of the
`Framingham cohort with unequivocally normal serum Cbl con-
`centrations in whom frank renal insufficiency was absent. There
`were 125 subjects with a cobalamin value > 369 pmol/L (500
`pg/mL), associated with a normal serum creatinine and a BUN
`< 10.7 nmol/L (< 30 mg/dL). Most of those in this subset that
`returned the questionnaire reported taking Cbl supplements. The
`highest MMA and Hcys concentrations in the 125 subjects were
`364 nmol/L and ! 8.0 #moFL, respectively, values < 3 SD above
`the mean in our younger control group.
`
`Correlations with serum metabolites
`
`As indicated by bivariate (ie, so-called "univariate")corre-
`lation coefficients unadjusted for confounding variables (Table
`
`TABLE 2
`Interpretation of findings in elderly subjects with elevated metabolite determinations and decreased serum concentrations of cobalamin (Cbl), folate,
`or both vitamins~
`
`Deficiency state
`
`Cbl (n = 41)
`Cbl (n = 3)
`Cbl _ folate (n = 21)
`Cbl and/or folate (n = 7)
`Folate (n = 2)
`
`Cbl
`(<258 pmol/L)
`
`Folate
`(< ! 1.3 nmol/L)
`
`t MMA
`
`1’ Hcys
`
`Number with
`creatinine
`
`+
`
`+
`+
`+
`-
`
`_
`
`-
`+
`
`+
`+
`
`+
`
`-
`+
`
`-
`-
`
`___2
`
`+
`__.3
`
`+
`+
`
`5
`i
`5
`2
`0
`
`~ MMA, methylmalonic acid; Hcys, total homocysteine; T creatinine, serum concentration > 106 tzmol/L (> 1.2 mg/dL) in females, > 124/~mol/
`L (> 1.4 mg/dL) in males; "t MMA, > 376 nmol/L; T Hcys > 21.3 ~mol/L. A Cbl concentration of 258 pmol/L = 350 pg/mL; a folate concentration
`
`of 11.3 nmoFL = 5.0 ng/mL.
`2 11 of 41 hadan T Hcys.
`3 lOof21 hadan t Hcys.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1105-0005
`
`

`
`COBALAMIN DEFICIENCY IN ELDERLY PEOPLE
`
`7
`
`3), the serum MMA concentration correlated most strongly with
`serum Cbl and also with various measures of renal function as
`well as with age. Only a weak correlation was apparent between
`MMA and serum folate. If the 29 subjects with Cbl values < 148
`pmol/L were excluded from the analysis, the correlation between
`serum Cbl and MMA was still good (r = -0.37, P < 0.001),
`but if the 222 persons with Cbl values < 258 pmolFL (< 350
`pg/mL) were excluded, the correlation was much weaker (r
`= -0.12).
`Serum Hcys correlated best with serum Cbl, folate, and cre-
`atinine and also correlated with serum uric acid, BUN, and age
`(Table 3). The correlation of serum Cbl and Hcys remained
`strong (r = -0.41) when subjects with Cbl concentrations < 148
`pmolFL were excluded, and remained significant (r = -0.28, P
`< 0.001) even when those with concentrations < 258 pmol/L
`were excluded, as did the correlation between serum folate and
`Hcys (r = -0.39). When the 16 subjects with folate values < 5.9
`nmol/L were excluded, the correlation between folate and Hcys
`values was only slightly weakened (r = -0.37). Even when the
`129 subjects with folate concentrations < 11.3 nmoUL were ex-
`cluded, it was lessened but still apparent (r = -0.28). On the
`other hand, removal of these 129 subjects had little effect on the
`correlation between Cbl and Hcys concentrations (r = -0.47).
`MMA and Hcys concentrations correlated strongly with each
`other, but neither metabolite correlated with the number of years
`of schooling or the amount of alcohol consumed. Multivariate
`stepwise-logistic-regression analysis was performed to assess the
`combined effects on each of the metabolite concentrations of the
`variables listed in Table 3. The serum Cbl concentration was
`found to be the strongest independent predictor of both MMA
`and Hcys concentrations (data not shown).
`
`Hematologic findings
`
`Blood counts were available from 501 of the 548 subjects.
`Anemia (hematocrit < 0.40 in men, < 0.35 in women) was pres-
`ent in 59 subjects (11.8%). In only eight subjects was the he-
`matocrit < 0.30. The anemia was typically normocytic (MCV
`80-100 fL) and in 20 was associated with an elevation of serum
`creatinine. In two anemic individuals the MCV was > 100 fL
`and in four it was < 80 fL. In an additional eight subjects the
`MCV was > I00 fL in the absence of anemia.
`In 3 of the 10 subjects with an MCV > 100 fL, Cbl was < 258
`pmol/L and MMA was increased (Table 4; subjects 1, 2, and 4)
`in the absence of anemia. Subjects 1 and 2 also had elevated
`Hcys values and decreased or low-normal folate concentrations.
`Cbl deficiency (with or without associated folate depletion) may
`have been responsible for the macrocytosis in these subjects. In
`an additional subject (Table 4; subject 3) with a high MCV, mild
`anemia, low serum folate, and a high Hcys concentration, folate
`deficiency may have been the cause of the macrocytic anemia.
`In 13 other subjects (Table 4; subjects 5-17), MCVs in the upper
`end of the normal range (95-99 fL) were associated with low or
`low-normal Cbl concentrations and elevations of one or both me-
`tabolites. It is possible that some of them had early hematologic
`abnormalities due to Cbl deficiency.
`Other than the subject with macrocytic anemia and laboratory
`data consistent with folate deficiency (Table 4; subject 3), there
`were 13 persons with serum folate values < 11.3 nmol/L who
`were anemic; Hcys was elevated in only 3 of them. In each of
`the three Cbl was < 258 pmol/L, serum creatinine was elevated,
`and the MCV was normal or low.
`
`The mean MCV and hematocrit of subjects with Cbl values
`< 258 pmol/L or with folate values < 11.3 nmol/L did not differ
`from those of subjects with higher values. However, the mean
`MCV of the subset of persons with a serum Cbl < 258 pmol/L
`and elevations of both metabolites (95.4 _ 8.9 fL, n = 17) was
`higher than that of the remaining subjects in the cohort (91.0
`+_ 5.4 fL, n = 484; P < 0.005), and the mean red blood cell
`count was lower (4.19 +_ 0.66 vs 4.50 _ 0.54 × 106 ceils/L,
`respectively; P < 0.025), although hematocrit values did not dif-
`fer (0.396 _ 0.052 vs 0.409 _ 0.048, respectively; P > 0.05).
`
`Antibodies to intrinsic factor
`
`Serum was available for testing for antibodies to intrinsic fac-
`tor from 58 of the 72 subjects with a Cbl concentration < 258
`pmolFL and an elevation of MMA and/or Hcys, including 19 with
`elevations of both metabolites, and 9 with Cbl concentrations
`< 148 pmol/L in the absence of such elevations. Antibodies to
`intrinsic factor were found in 1 of the 67 subjects (a 79-y-old
`man with a hematocrit of 0.42, an MCV of 93 fL, serum Cbl of
`118 pmol/L, MMA of 1301 nmolFL, and Hcys of 33.8 btmol/L).
`
`Vitamin supplements
`
`Of the 401 subjects responding to the questionnaire, 115
`(28.7%) and 81 (20.2%) reported taking oral vitamin supple-
`ments containing Cbl and folate, respectively. The median daily
`doses were 6 #g Cbl and 400/.zg folic acid. Nineteen (I 1.9%) of
`160 subjects with Cbl values < 258 pmoi/L reported taking sup-
`plements containing Cbl, in contrast with 96 (39.8%) of 241 with
`higher values (X2 = 52, P < 0.001). Cbi supplements were taken
`by 8 (14.2%) of 56 subjects with MMA elevations, in contrast
`with 107 (31.0%) of 345 without elevations (P < 0.02). Three
`of i 0 subjects who had high MMA concentrations and Cbl values
`< 148 pmol/L reported ingesting Cbl supplements. Supplements
`containing folic acid were taken by 65 (41.9%) of 155 persons
`with serum folate concentrations > 20 nmol/L, compared with
`16 (6.5%) of 246 with lower folate values (P < 0.001).
`After log transformation of the data, the reported intake of the
`respective vitamins correlated with serum vitamin concentrations
`(r = 0.39 for cobalamin and 0.43 for folate); the correlations with
`serum metabolite concentrations, however, were not impressive
`(for Cbl intake with MMA, r = -0.15 and with Hcys, r = -0.26;
`for folic acid intake and Hcys, r = -0.25).
`
`Discussion
`
`Two major findings were demonstrated by this study: 1) a
`strikingly high prevalence of metabolically significant Cbl defi-
`ciency was found in the large, well-defined ambulatory elderly
`population of the Framingham community, and 2) the majority
`of subjects with metabolic evidence of deficiency had serum Cbl
`concentrations within the conventionally defined normal range.
`Of the 548 members of the cohort, 65 (11.9%) had elevations of
`one or both metabolites and a serum Cbl concentration < 258
`pmol/L (Table 2), findings strongly suggestive of Cbl deficiency
`(21, 23, 28-32). An additional seven subjects (1.3%) may have
`been deficient in Cbi (Table 2, line 4). The prevalence of defi-
`ciency may have been even greater, because an additional 35
`subjects (6.4%) with serum Cbl concentrations < 258 pmol/L
`had MMA values between 2 and 3 SD above the mean in young
`control subjects in association with normal serum creatinine con-
`
`0
`
`0
`
`m
`
`©
`
`©
`
`©
`
`o
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1105-0006
`
`

`
`LINDENBAUM ET AL
`
`TABLE 3
`Pearson correlation coefficients (r) between serum metabolites (MMA
`and Hcys) and other variables in elderly subjects/
`
`Serum cobalamin
`Serum folate
`Serum creatinine
`Creatinine clearance
`Serum urea nitrogen
`Serum uric acid
`Age
`Serum Hcys
`
`MMA
`
`-0.46
`-0.12
`0.30
`-0.30
`0.24
`0.17
`0.25
`0.64
`
`Hcys
`
`-0.50
`-0.42
`0.38
`-0.28
`0.25
`0.33
`0.27
`--
`
`J P values for all correlations were < 0.001, except for the correlation
`between serum folate and MMA (P < 0.01). For all variables except
`age, creatinine clearance, and uric acid, data were log transformed before
`analysis. MMA, methylmalonic acid; Hcys, total homocysteine.
`
`centrations, and a few subjects with serum Cbl values -> 258
`pmol/L may have been deficient (16, 31).
`The impressively high prevalence of metabolic evidence of
`lack of Cbl in a substantial cross-section of the Framingham el-
`derly community is consistent with the findings of four recently
`reported series of consecutively studied geriatric outpatients
`seeking care in clinic or private office settings in whom serum
`metabolites were measured (16, 44-46). As in many previous
`reports, the hematologic changes typical of megaloblastic anemia
`(47) were absent in the great majority of subjects with evidence
`of deficiency, although modest abnormalities were demonstrable
`in a relatively small subset.
`
`We chose a serum Cbl concentration of 258 pmoi/L (350 pg/
`mL) as a cutoff point for suspecting Cbi deficiency, based on the
`overrepresentation of high serum MMA values in members of
`the cohort with Cbl concentrations below this number (Fig 3).
`Also, the degree of elevation of metabolite concentrations in sub-
`jects with Cbl determinations of 74-147 pmol/L was similar to
`that seen in subjects with values of 148-258 pmol/L (Fig 2).
`Indeed, the correlation between serum Cbi and each of the me-
`tabolite concentrations remained robust even when members of
`the cohort with serum vitamin values < 148 pmoFL were ex-
`cluded from the analysis. Other investigators have also noted a
`negative correlation between serum Cbi values within the normal
`range and serum Hcys (25, 48-51).
`There is other evidence to indicate that a conventional lower
`limit of normal for serum Cbl, as determined by subtracting 2
`SDs from the mean of a young control population, will fail to
`detect a significant number of patients who are depleted of the
`vitamin. At least 5% of patients with unequivocal clinical evi-
`dence of Cbl were found to have serum Cbl determinations in
`t