- Email: [email protected]
Sensitivity of serum methylmalonic acid and total homocysteine determinations for diagnosing cobalamin and folate deficiencies
Sensitivity of Serum Methylmalonic Acid and Total Homocysteine Determinations for Diagnosing Cobalamin and Folate Deficiencies DAVIDG. SAVAGE,M.D., JOHN LINDENBAUM,M.D., NewYork,NewYork,S~~~~ P. STABLER,M.D., ROBERTH. ALLEN, M.D., Denver,Colorado PURPOSE: Patients with cobalamin (vitamin B12) deficiency usually lack many of the classic features of severe megaloblastic anemia; because of the low diagnostic specificity of decreased serum cobalamin levels, demonstrating the deficiency unequivocally is often difficult. We examined the sensitivity of measuring serum concentrations of methylmalonic acid and total homocysteine for diagnosing patients with clear-cut cobalamin deficiency and compared the results with those of patients with clear-cut folate deficiency. PATIENTS AND METHODS: Serum metabolites were measured for all patients seen from 1982 to 1989 at two university hospitals who met the criteria for cobalamin and folate deficiency states and for such patients seen from 1968 to 1981 from whom stored sera were available. In all, 406 patients had 434 episodes of cobalamin deficiency and 119 patients had 123 episodes of folate deficiency. Criteria for deficiency states included serum vitamin levels, hematologic and neurologic findings, and responses to therapy. Responses were documented in 97% of cobalamin-deficient patients and 76% of folatedeficient patients. Metabolite levels were measured by modified techniques using capillarygas chromatography and mass spectrometry. RESULTS: Most of the cobalamin-deficient patients had underlying pernicious anemia; two thirds were blacks or Latinos. Hematocrits were normal in 28% and mean cell volumes in 17%. Of the 434 episodes of cobalamin deficiency, 98.4% of serum methylmalonic acid levels and 95.9% of serum homocysteine levels were elevated (greater than 3 standard deviations above the mean in normal subjects). Only one patient had normal levels of both metabolites. Serum From the Departments of Medicine, Columbia University, College of Physicians and Sugeons, Columbia-Presbyterian Medical Center, and Harlem Hospital Center, New York City (DGS, JL); Department of Medicine (SPS, RHA) and Department of Biochemistry, Biophysics and Genetics (RHA). Universitv of Colorado Health Sciences Center. Denver. Colorado. This work was supported in part by Department of Health and Human Services Research Grants (DK-21365 and AG-09834) from the National Institutes of Health. Requests for reprints should be addressed to Dr. John Lindenbaum, Department of Medicine, Columbia-Presbyterian Medical Center, 630 West 168th Street, New York, New York 10032. Manuscript submitted December 29, 1992, and accepted in revised form April 8, 1993.
March
homocysteine levels were increased in 91% of the 123 episodes of folate deficiency. Methylmalonic acid was elevated in 12.2% of the folate-deficient patients; in all but one, the elevation was attributable to renal insufficiency or hypovolemia. CONCLUSIONS: For the cobalamin-deficient patients, measuring serum metabolite concentrations proved to be a highly sensitive test of deficiency. We conclude that normal levels of both methylmalonic acid and total homocysteine rule out clinically significant cobalamin deficiency with virtual certainty. obalamin (vitamin Blz) deficiency is common. Early diagnosis and treatment can prevent severe anemia or potentially irreversible damage to the nervous system from developing, but the majority of patients lack the textbook features of advanced de& ciency of the vitamin [l-4]. Serum cobalamin concentration is widely used as a screening test for deficiency but has major limitations. Its sensitivity is less than perfect [5,6], and many workers [4,7-121 have found that a substantial proportion-perhaps 25% to 50% [4,12]of patients with low serum cobalamin levels do not appear to be deficient in the vitamin or to have underlying disorders predisposing them to deficiency. The metabolites methylmalonic acid and homocysteine accumulate when the two mammalian cobalamin-dependent enzymatic reactions are impaired [ 13,141. We found that serum levels of both metabolites are markedly elevated (more than three standard deviations [SD] above the mean in healthy control subjects) in most deficient patients, suggesting that these tests are a useful and convenient way to establish a diagnosis of cobalamin deficiency [4&L-15]. The assays were modified in early 1988 to further increase their specificity and precision [ 16,171. We report here using the modified techniques to assay more than 400 patients with clinical evidence of cobalamin deficiency. To our knowledge, this series, accumulated over 21 years, is the largest group of patients with well-documented cobalamin deficiency reported in the literature. For comparison, we assayed a series of 119 patients with folate deficiency. In previous studies involving relatively small numbers of folate-deficient patients, total homocysteine concentrations were elevated in most cases [14,18,19].
C
1994
The American
Journal
of Medicine
Volume
96
239
TABLE I Criteria Serum
TABLE III
for Cobalamin Cobalamin
Deficiency
< 200 pg/mL
Diagnostic Marrow +/or Blood Smear +
in 434 Episodes
in 401 Patients
Criteria
Response to Cbl Rx*
No. of Episodes 309 13 107 5
N+A + +
rTA 0 Total
% 71.2 3.0 24.6 1.2
434
100
NA = not available; Cbl = cobalamin; Rx = therapy. *See Table II for definition of responses.
TABLE II Responses Deficiency
to Cobalamin Treatment in 409 Patients
in 434 Episodes
Decrease in mean cell volume > 5 fL Increase in hematocrit 2 0.05 Resolution of thrombocytopenia Resolution of leukopenia Improvement in neuropsychiatric abnormalities* More than one of above responses to treatment Other responsest No documentation of response
of Cobalamin
No. of Episodes 369 286 110 50 112 325 9 13
.g 85.0 65.9 25.3 11.5 25.8 74.9 2.1 3.0
*Clear-cut improvement in neuropsychiatric abnormalities within the first three months of treatment. This was the only response to treatment in a single episode. TOnly one of the following responses was documented: reticulocytosis (five patients); correction of neutrophil hypersegmentation (three patients); and resolution of atrophic glossitis (one patient).
PATIENTS AND METHODS Patients All cases of cobalamin and folate deficiency seen at Harlem Hospital Center and Columbia-Presbyterian Medical Center between July 1968 and June 1989 were reviewed. The majority of the patients were evaluated in consultation with one of the coauthors as part of prospective studies of megaloblastic anemias, and the hematologic and neurologic findings have been the subject of previous reports [4,13-15,20-261. Methylmalonic acid and total homocysteine levels were measured in all patients seen between 1968 and 1981 who met the study’s criteria for cobalamin and folate deficiency and from whom serum stored at -20°C was available. From 1982 to 1989, serum metabolite levels were measured in each of 267 consecutively seen patients who met the cirteria for cobalamin deficiency (219 patients) or folate deficiency (48 patients). Methods When available, smears of peripheral blood and bone marrow from cobalamin- and folate-deficient patients were coded and interspersed with those of patients without evidence of deficiency for blinded review by one of the authors. Hypersegmentation in 240
March
1994
in 119 Patients No. of Episodes Diagnostic marrow or blood smear, or both 123 Serum folate i 4 ng/mL* 123 Serum cobalamin > 300 pg/mLT 123 Underlying disorder associated with folate deficiency 123 Documented response to folic acid 94
and
The American
Journal
of Medicine
Volume
for Folate Deficiency
*Serum folate < 2.1 ng/mL TSerum cobalamin 301 to 10, 401 to 500 pg/mL in 1,000 pg/mL in 26. None before study.
in 123 Episodes
?j 100 100 100 100 76.4
in 99 episodes and 2.1 to 3.9 ng/mL in 24. 359 pg/mL in 7 episodes, 351 to 400 pg/mL in 20, 501 to 1,000 pg/mL in 60, and greater than of the patients had been treated with vitamin EP*
blood smears was defined as more than five cells with five lobes per 100 neutrophils or any cells with six or more lobes. Serum cobalamin levels were determined by radioassay using purified intrinsic factor (Quantaphase, Bio-Rad Laboratories, Richmond, CA) or by microbiologic assay with LactobaciUus leichmannii [27]. Serum folate concentrations were measured by milk-binder radioassays or by microbiologic assay with Lactobacillus casei [28]. Serum concentrations of methylmalonic acid 1291 and total homocysteine [30] were measured by modified [ 16,171 techniques using capillary-gas chromatography and mass spectrometry. The normal ranges (calculated as the mean & 3 SD after log transformation to correct for skewing toward higher values) are 53 to 376 nmol/L for methylmalonic acid and 4.1 to 21.3 urnol/L for total homocysteine [ 161. For 160 of the patients with cobalamin deficiency and 19 with folate deficiency, serum metabolite levels were measured by methods used before our modifications were made, and the results were reported earlier [4,13-151. Standard statistical methods, including Student’s t-test and the &i-squared test with Yates’ correction for continuity [31], were used to analyze the data
RESULTS Criteria of Deficiency Serum methylmalonic acid and total homocysteine levels were measured in 406 patients who had 434 episodes of cobalamin deficiency; 25 patients had one or more recurrent episodes of deficiency because they discontinued maintenance treatment with the vitamin. The diagnostic criteria for including cobalamindeficient patients in this series are summarized in Table I. All patients had low serum cobalamin levels. Morphologic abnormalities (megaloblastic bone marrow or hypersegmented neutrophils, or both, with or without macroovalocytes, on blood smear) were found in 322 episodes; clear-cut responses to cobalamin therapy (as defined in Table II) were documented for 309 of the 322 episodes. Although marrow and blood films were not available for 107 episodes before therapy began, clear-cut responses were doc96
SERUM METABOLITES IN COBALAMIN AND FOLATE DEFICIENCY / SAVAGE ET AL TABLE IV Etiology
of Cobalamin Etiology
TABLE V and Folate
Deficiencies
in 529 Patients Patients
Cobalamin deficiency Pernicious anemia Proven Probable* Tropical sprue Proven Probablet Gastrectomy lleal disease or resection, or both Gastric and ileal resections Jejunal diverticula Vegetarianism Food cobalamin malabsorption Not established Total
234 76 11 28 13 13 1 6 2 3 19
‘One or more of the following: cobalamin malabsorption Schilling test, achlorhydria, and autoimmune thyroid disease.
37.2 62.8
70/l 19 49/119
58.8 41.2
Ethnicity Black White Latin0 Other
168/406 135/406 101/406 2/406
41.3 33.3 24.9 0.5
105/119 9/l 19 5/119 o/119
88.2 7.6 4.2 0.0
59/434
13.6
107/123
87.0
Anemia
X3/434
72.1
123/123
100.0
Macrocytosis MCV > 100 fL MCV> IlOfL
334/402 224/402
83.1 55.7
68/91 50/91
74.7 54.9
73/433
16.9
22/123
17.9
136/384
35.4
78/l 18
66.1
Pancytopenia
48/384
12.5
21/118
17.8
LDH > 1,000 U/L
96/330
29.1
24/83
28.9
11 l/434
25.6
27/123
22.0
156/434
35.9
36/123
29.3
ThrombocytopeniaS
Atrophic glossitis Neuropsychiatric
in part 1 of the
in a person who had lived in the tropics
151/406 255/406
Leukopeniat
119
(intestinal
unrented; all but one of these episodes showed at least one of the hematologic responses listed in Table II. In five episodes showing a therapeutic response, pretreatment blood smears were normal or nondiagnostic (marrow aspirates were not obtained). Of the 434 episodes, 421 (97%) responses to treatment with cyanocobalamin were documented (Table ID. Serum metabolites were measured in 123 episodes of folate deficiency diagnosed in 119 patients (four patients had two episodes). Our definition of folate deficiency was based on the clinical, morphologic, and biochemical criteria summarized in Table III. Hematologic responses to treatment with folic acid were documented for 94 episodes; adequate hematologic follow-up data were not obtained for the others. Patients were considered folate or cobalamin deficient based solely on the stated criteria and not on the results of the metabolite assays. The majority of the patients (86%) were studied before 1986, when the metabolite assays became available, and the percent of patients with each deficiency with elevated metabloite levels was similar for patients who were evaluated before and after 1986.
CAUSES OF DEFICIENCY STATES Diagnostic studies were adequate to determine the etiology of cobalamin deficiency for most patients March
Folate Deficiency No./Totalt %
Sex Male Female
Alcoholism
103 11 1 1 1 1 1
Total
Cobalamin Deficiency No./TotalT %
Characteristics*
406
Folic acid deficiency Alcoholism Malnutrition Malabsorption syndrome ? Oral contraceptives Pregnancy Sickle cell anemia Tropical sprue
TMalabsorption syndrome biopsy not performed).
Clinical Characteristics of 434 Episodes of Cobalamin Deficiency in 406 Patients and 123 Episodes of Folate Deficiency in 119 Patients
syndromes
MCV = mean corpuscular volume; LDH = lactate dehydrogenase. *Mean age 49 * 17 years in folate-deficient group and 66 f 18 years in cobalamin-deficient patients. tTotal number of cases for which information was available; denominator = number of patients, for sex and ethnic background, and number of episodes for other items. SLeukopenia = leukocyte count less than 4,OOO/pL for white and Latin0 patients and less than 3,OOO/pL for blacks; thrombocytopenia = platelet count less than 150,000/$.. .$Consistent with those syndromes seen in cobalamin deficiency f261. Peripheral neuropathy, ataxia, and cerebral dysfunction were often seen in folate-deficient alcoholics.
(Table IV). Pernicious anemia, demonstrated by serial Schilling tests or serum antibody to intrinsic factor, was established for 57.6% and considered likely for another 18.7%. The serum folate level was less than 2.1 ng/mL in 17 patients and between 2.1 ng/mL and 3.9 ng/mL in 57 patients in the cobalamindeficiency group. All 74 patients had one or more fmdings of serum antibodies to intrinsic factor, malabsorption of vitamin BLzin Schilling tests, or neurologic dysfunction responsive to cobalamin treatment. Of these 74 patients, 19 had proven or probable tropical sprue, and it is likely that some had folate deficiency along with cobalamin deficiency. The 19 cobalamindeficient patients for whom diagnostic evaluation was inadequate to determine the etiology of the deficiency (Table IV) had serum folate levels greater than 4 ng/mL in all cases. Alcoholism caused 87% of the episodes of folate deficiency (Table IV).
Characteristics of Episodes The clinical features of the cobalamin- and folatedeficient patients are summarized in Table V. The 1994
The American
Journal
of Medicine
Volume
96
241
TABLE VI Serum Concentrations of Methylmalonic Homocysteine in Episodes of Cobalamin Without Anemia
Acid Deficiency
and Total With and
AnemiaPresent*AnemiaAbsent All Episodes No. % No. % No. % Total numberof episodes 313 121 434 Episodeswith ElevatedsMMA 308 98.4 119 98.3 427 98.4 ElevatedsHCYS 306 97.8t 110 90.9t 416 95.9 Both MMA, HCYS elevated MMA alone elevated HCYS alone elevated MMA, HCYS both normal
302 6 4 1
96.5t 1.9 1.3 0.3
108 11 2 0
89.3$ 9.2 1.7 0
410 17 6 1
94.5 3.9 1.4 0.2
MMA= methylmalonicacid; HCYS= total homocysteine.
. 50
100
150
200
250
*Hematocrit corpuscular anemia was tp <0.005. $p
300 ‘W%5
HOMOCYSTEINE, pmol/L
.
less than 40% in males; less than 35% in females. The mean volume was normal (80 to 100 fL) in 9.9% of the episodes when present and in 33.3% of those when anemia was absent.
3 SD above the mean in normal control subjects.
TABLE VII
.
I
Serum Metabolite Concentrations in Anemia Caused by Cobalamin or Folate Deficiency Cobalamin Deficiency Folate Deficiency No. 96 No. %
.
Total no. of episodes*
297
Elevated MMA
292
98.3
98
Elevated HCYS
290
97.6
88
89.8
Both MMA,HCYSelevated
4.1
4t
4.1
286
96.2
4
Elevated MMA, normal HCYS
6
2.0
0
Elevated HCYS. normal MMA
4
1.3
84
85.7
MMA,HCYSboth normal
1
0.3
10
10.2
0
MMA= methylmalonicacid; HCYS= total homocysteine. *Sixteen patients with cobalamin deficiency and 25 with folate deficiency elevated serum creatinine levels were excluded from the analysis.
with
t.Severe volume depletion present in three episodes. .-
a--,--------------
I
50
100 HOMOCYSTEINE,
L
150 FmoliL
I
200 220
Figure 1 Top. Serum concentrations of methylmalonic acid and total homocysteine in 313 episodes of megaloblastic anemia due to cobalamin deficiency. Dashed lines indicate three standard deviations above the mean for normal controls for each metabolite; hatched area includes values for both metabolites less than three standard deviations above the mean for normal controls; open circles indicate 16 cobalamin deficiency episodes associated with anemia with elevated serum creatinine. Figure 1 Bottom. Serum cdncentrations of methylmalonic acid and total homocysteine in 121 episodes of cobalamin deficiency in which the hematocrit was normal (greater than or equal to 40% in men, greater than or equal to 35% in women). Dashed lines indicate three standard deviations above the mean for normal controls for each metabolite; hatched area includes values for both metabolites less than three standard deviations above the mean for normal controls. 242
March 1994
The American Journal of Medicine
substantial numbers of black and Latin0 patients with cobalamin deficiency reflect the distribution of ethnic groups in the catchment areas of the two hospitals and are consistent with previous reports of a high prevalence of pernicious anemia in these populations [32-34]. The higher proportion of black patients and anemic patients in the folate-deficient group came from including several series of alcoholics with anemia studied at Harlem Hospital Center [23,25]. The mean corpuscular volume (MCV) was normal in 17% of cobaknnin-deficient patients tid in 25% of folate-deficient patients. Only a minority of patients with either deficiency state had leukopenia, pancytopenia, atrophic glossitis, or marked elevations of serum lactate dehydrogenase. More than a third of the cobalamindeficient group has neuropsychiatric symptoms and signs (Table V) compatible with that vitamin deficiency [26]. Peripheral neuropathy, ataxia, and cerebral dysfunction were frequently associated with alcoholism in the folatedeficient patients.
Volume 96
SERUM
METABOLITES
Serum Metabolites in Cobalamin Deficiency Serum methylmalonic acid levels were markedly increased (more than 3 SD above the mean in normal controls) in 427 (98.4%) of the 434 episodes of cobalamin deficiency (Figure 1, Table VI); serum total homocysteine concentrations were markedly increased (more than 3 SD above the mean in normal controls) in 416 (95.9%). In 410 episodes (94.50/o),both metabolite levels were markedly elevated. In only one cobalamin-deficient patient (0.2%) were both metabolites normal. The hematocrit was normal (Figure 1 Bottom; Table VI) in 121 of the 434 episodes of cobalamin deficiency. Nonetheless, serum methylmalonic acid values were elevated in 98.3% of the episodes without anemia In contrast, homocysteine levels were elevated in 90.9% of the nonanemic patients compared with 97.8% of anemic patients (Table VI, chi squared = 8.66, p ~0.005). Thus serum methyhnalonic acid levels were more sensitive for identifying nonanemic cobalamindeficient patients than serum homocysteine levels (p ~0.005). Both metabolites were elevated in 89.3% of the episodes in which the hematocrit was normal. Mean homocysteine levels were significantly higher in anemic than nonanemic patients (89.4 + 55.0 umol/L versus 60.2 t- 41.3 pmol/L, p0.05). The clinical, hematologic, and neurologic features of the 18 episodes with normal serum homocysteine levels did not differ from those of the 416 with elevated levels, except that anemia was generally less severe when they were normal (mean hematocrit, 37.2% versus 28.3%, p
IN COBALAMIN
IO
20
AND
50
FOLATE
DEFICIENCY
100
150
HOMOCYSTEINE,
/ SAVAGE
200
ET AL
250
pmol/L
Figure 2. Serum concentrations of methylmalonic acid and total homocysteine in 123 episodes of megaloblastic anemia due to folate deficiency. Open circles indicate patients with elevated serum creatinine concentrations; boxes indicate three patients with severe hypovolemia and normal creatinine levels. Dashed lines indicate three standard deviations above the mean for normal controls. (Note that the scale for methylmalonic acid values differs from that in Figure 1.)
obtained during the most recent 5 years of the study, 1984 to 1989 (data not shown). In a series of 86 consecutively studied patients with cobalamin deficiency [4] tested by earlier assay methods, 12 (14%) did not have elevated methylmalonic acid and 13 (15.1%) did not have increased homocysteine. Sera were available from most of these patients for assay by the modified techniques used in this study. Seven of 10 patients with methylmalonic acid levels less than 3 SD above the normal mean using the older method had elevated concentrations using the current assay system. Of 11 previously normal homocysteine values, four were increased when reassayed by the modified method. Serum Metabolites in Folate Deficiency Serum homocysteine levels were markedly elevated in 112 (91.0%) of the 123 episodes of folate deficiency (Figure 2). The 11 patients with normal levels did not differ from the others in severity of anemia, megaloblastic morphologic changes, methylmalonic acid elevation, time in hospital before study, reticulocyte counts, or serum levels of folate, cobalamin, creatinine, bilirubin, or albumin (data not shown). Serum methylmalonic acid was increased in 15 (12.2%) of the 123 episodes of folate deficiency (Figure 2); however, 14 of these patients had renal dysfunction (11 with serum creatinine elevation and 3 with clinical evidence of severe volume depletion). 1994
The American
Journal
of Medicine
Volume
96
243
Excluding these 14 patients leaves serum methylmalonic acid elevation in only one episode of folate deficiency; and in that patient, the increase was modest (398 &I/L). Serum Metabolites in Cobalamin Versus Folate Deficiency The serum metabolite levels of the folate-deficient patients, all of whom were anemic, were compared with those of the cobalamin-deficient patients with anemia (Table VII). Patients with elevated serum creatinine values (16 with cobalamm deficiency and 25 with folate deficiency) were excluded from the analysis. Serum methylmalonic acid was markedly elevated in 98.3% of the 297 cobalamin deficiency episodes and in only 4.1% of the 98 folate deficiency episodes (chi squared = 343, p
March
1994
The American
Journal
of Medicine
Volume
COMMENTS Concentrations of methylmalonic acid and homocysteine are stable indefinitely in serum stored at -20°C [13,14], and we found no trend toward lower values in serum that had been repeatedly thawed and frozen over a period as long as 20 years. Hence, the stability of these compounds allowed us to test their diagnostic usefulness for a large series of patients with well-documented cobalamin and folate deficiency that would have required decades to assemble prospectively. We included all patients with these deficiencies seen at two New York City hospitals between 1968 and 1989 from whom serum was available, after excluding patients with low or borderline levels of both vitamins when it was uncertain which deficiency they had. The results point to the high sensitivity of serum methylmalonic acid and homocysteine concentrations, using the modified assay methods, for diagnosing patients with clinically significant cobalamin deficiency. One or the other metabolite level was markedly elevated (greater than 3 SD above the mean in normal controls) in all but one of the 434 episodes (Table VI, Figure l), giving a combined sensitivity of 99.8% for the metabolites for detecting deficiency of the vitamin. While methylmalonic acid and homocysteine concentrations were both increased in approximately 98% of cobalamin-deficient patients with megaloblastic anemia, the methylmalonic acid level was more consistently elevated in nonanemic patients (Table VI). Using the modified techniques, methylmalonic acid determinations also had greater sensitivity for detecting mild or early cobalamin deficiency in patients with pernicious anemia receiving infrequent maintenance therapy with cobalamin [6] as well as in elderly patients with low-normal serum cobalamin levels [35]. Moelby and colleagues also found serum methylmalonic acid elevated in 30 of 31 patients with cobalamin deficiency [ 121. In spite of the generally high sensitivity of serum methylmalonic acid determinations, concentrations were normal in 7 of 434 episodes of cobalamin deficiency (Table VI); however, the homocysteine level was increased in six of the seven episodes, indicating that the two metabolite determinations combined improve the rate of detecting cobalamin deficiency over that of methylmalonic acid alone. The reason patients occasionally fail to show a methylmalonic acid elevation is not clear, although gut flora alterations that disturb propionate production by enteric bacteria may play a role [61. We have included in the cobalamin-deficient group 74 patients whose serum folate levels were also low or at the lower end of the normal range. Most of these patients had underlying pernicious anemia, documented by serum antibodies to intrinsic factor, or 96
SERUM METABOLITES IN COBALAMIN AND FOLATE DEFICIENCY / SAVAGE ET AL
cobalamin malabsorption corrected by intrinsic factor in Shilling tests performed many weeks after correcting the deficiency state. In some of the patients with tropical sprue, however, folate deficiency may have coexisted with cobalamin deficiency, or conceivably, may have been the primary deficiency. Serum methylmalonic acid was elevated in 72 and serum homocysteine in all of the 74 patients considered cobalamin deficient who had low or low-normal folate values. Excluding these patients from the analysis would not significantly change any of the percentages listed in Table K-in fact, the percentage of patients with cobalamin deficiency and increased methylmalonic acid level would be even higher. Patients with folate deficiency would be expected to have a block in the conversion of homocysteine to methionine [14]; and indeed, serum homocysteine was markedly increased in 91% of the 123 episodes of folate deficiency, confiig earlier reports of smaller numbers of patients [ 14,18,19]. The pattern of elevated homocysteine associated with normal methylmalonic acid in a patient with a megaloblastic anemia was much more commonly associated with folate deficiency than with cobalamin deficiency: 84 of 88, or 95% of such episodes (Table VII) were caused by folate deficiency. Serum methylmalonic acid would not be expected to accumulate as a result of lack of folate; however, it was increased in 15 (12%) of the 123 episodes of folate deficiency. Methylmalonic acid is excreted in the urine. Eleven of the 15 patients with folate deficiency and elevated serum methylmalonic acid had renal function impairment severe enough to increase serum creatinine; three other patients had marked intravascular volume depletion, which may have led to renal retention of methylmalonic acid; for only one folatedeficient patient was there no explanation for the elevation. Therefore, impaired renal function appears to be the principal cause of a false-positive increase in serum methylmalonic acid in patients who do not appear to be cobalamin deficient. For the majority of patients undergoing evaluation for possible cobalamin deficiency, available clinical and laboratory data on renal function and intravascular volume status will indicate whether changes in these conditions may be responsible for an elevated methylmalonic acid. Furthermore, renal dysfunction by itself appears to cause only a modest increment in the serum methylmalonic acid level (Figure 2); for most patients with cobalamin deficiency, the methylmalonic acid concentration exceeds that attributable to renal insufficiency alone (compare Figures 1 and 2). We are currently studying a large series of patients with renal disorders and no cobalamin or folate deficiency to determine the upper limit of the methylmalonic acid level caused by disturbances in kidney function. March
Although total homocysteine is elevated in both cobalamin and folate deficiency, making the correct diagnosis is important since treating a cobalamindeficient patient with folic acid alone will not prevent or correct the neuropsychiatric abnormalities caused by the deficiency, even though partial or even complete hematologic responses to folic acid often occur 136,371. In addition, elevated homocysteine will respond only to treatment with the proper vitamin this is true even when cobalamin-deficient patients have hematologic responses to folic acid [ 161. Normalizing an elevated homocysteine level may be important in itself since a number of recent studies have shown that elevated levels are strongly correlated with coronary artery, cerebrovascular, and peripheral vascular disease [38-40]. Some alcoholic patients with serum folate concentrations between 2.1 and 4.0 ng/mL (i.e., at the lower end of the normal range) were included in the folatedeficient group: all had megaloblastic anemia, unequivocally normal serum cobalamin values (Table III), and clear-cut responses to folic acid treatment, and all were considered highly likely to be folate deficient. A number of previous studies of alcoholic patients have noted normal serum folate values (commonly in the low-normal range) in 25% to 55% of patients with florid evidence of folate deficiency [25,4143]. Our findings from this large series of patients indicate that serum metabolite levels are highly sensitive tests for diagnosing cobalamin deficiency-normal levels of both methylmalonic acid and homocysteme rule out a deficiency with virtual certainty (Table VI). Furthermore, an elevated serum level of methylmalonic acid (or of both methylmalonic acid and homocysteine) is much more specific for cobalamin deficiency than is a low serum cobalamin level, which is frequently seen without any other evidence of deficiency [4,7-121. The only important limitation to the specificity of an increased serum methylmalonic acid concentration is underlying renal dysfunction, which the great majority of cobalamin-deficient patients do not have. Inborn errors of metabolism may lead to elevated serum homocysteine or methylmalonic acid; based on our experience, other causes of falsepositive serum metabolite elevations appear to be quite rare. Measurements of serum levels of these metabolites, available through commercial laboratories, appear to be highly useful ancillary tests for interpreting low or low-normal serum cobalamin concentrations, particularly since many, if not most, deficient patients lack the full-blown textbook manifestations of megaloblastic anemia (Table V) [14]. The availability of metabolite testing may obviate the need for Schilling tests or for prolonged therapeutic trials that may yield 1994
The American
Journal
of Medicine
Volume
96
246
SERUM METABOLITES IN COBALAMIN AND FOLATE DEFICIENCY / sAVAGE ET AL l__l~-.._.
equivocal results for many patients. Although the Schilling test measures cobalamin absowtion rather than deficiency, it is often used as a secondary test when a diagnosis of cobalamin deficiency is uncertain, on the assumption that deficiency is more likely if the Schilling test is abnormal. The metabolites provide more direct evidence of deficiency of the vitamin. Further studies will be required to determine whether metabolite level measurements should serve mainly as backup tests after serum cobalamin and folate values are obtained for patients with suspected deficiencies, or whether they should replace serum vitamin concentrations as the prim&y screening tests for these disorders.
REFERENCES 1. Blundell EL, Matthews JH, Allen SM, Middleton AM, Morris JE, Wickramasinghe SN. Importance of low serum vitamin B12 and r@d cell folate concentrations in elderly hospital inpatients. J Clin Path 1985; 38: 1179-1184. 2. Carmel R. Pernicious anemia-the expected findings of very low serum cobalamin levels, anemia, and macrocytosis are often lacking. Arch Intern Med 1988; 148: 1712-4. 3. Matthews JH, Clark DM, Abrahamson GM. Effect of therapy with vitamin Bli and folic acid on elderly patients with low concentrations of serum vitamin 812 or erythrocyte folate but normal blood counts. Acta Haemat 1988; 79: 84-87. 4. Stabler SP, Allen RH, Savage DG, Lindenbaum J. Clinical spectrum ‘and diagnosis of cobalamin deficiency. Blood 1990; 76: 871-81. 5. Magnus FM. Cobalamin and unsaturated transcobalamin values in pernicious anaemia: relation to treatment. Stand J Haematol 1986; 36: 457-65. 6. Lindenbaum J, Savage D, Stabler SP, Allen RH. Diagnosis of cobalamin deficiency. II. Relative sensltlvrtles of serum cobalamin, methylmalonic acid and total homocysteine concentrations. Am J Hematol 1990; 34: 99-107. 7. Bain B, Broom GN, Woodside J, Litwinczuk RA, Wickramasinghe SN. Assessment of a radioisotopic assay for vitamin B12 using Bn intrinsic factor preparation with R proteins blocked by vitamin B12 analogues. J Clin Path 1982; 35: 1110-3. 8. Schilling RF, Fairbanks VF, Miller R, Schmitt K, Smith MJ. Improved vitamin B12 assays, a report on two commercial kits. Clin Chem 1983; 29: 582-3. 9. Brynskov J, Andersen K, Gimsing P, Hippe E. False low serum vitamin B12 values with radiodilution assays using blocked R-binders. Lancet 1983; 1: 1104-5. : 10. Cooper BA, Fehedy V, Blanshay P. Recognition of deficiency of vitamin Bi2 using measurement of serum concentration. J Lab Clin Med 1986; 107: ‘: 447-52. / 11. Thompson WG, Cassino C, Babitz L, et al. Hypersegmented neutrophils !and vitamin 812 deficiency. Acta Haemat 1989; 81: 186-191. ‘;12. Moelby L, Rasmussen K, Jensen MK, Pedersen KO. The relationship between clinically confirmed cobalamin deficiency and serum methylmalonic iacid. J Int Med 1990; 228: 373-378. ‘13. Stabler SP, Marcell PD, Podell ER, Allen RH, Lindenbaum J. Assay of methylmalonic acid in the serum of patients with cobalamin deficiency using capillary gas chromatography-mass spectrometry. J Clin Invest 1986; 77: 1606-12. 14. Stabler SP, Marcell PD, Podell ER, Allen RH, Savage D, Lindenbaum J. Elevation of total homocysteine in the serum of patients with cobalamin or folate deficiency detected by capillary gas chromatography-mass spectrometry. J Clin Invest 1988; 81: 466-74. 15. Lindenbaum J, Healton EB, Savage D, et al. Frequency of neuropsychiatric disorders caused by cobalamin deficiency in the absence of anemia or marcocytosis. New Engl J Med 1988; 318: 1720-8. 16. Allen RH, Stabler SP, Savage D, Lindenbaum J. Diagnosis of cobalamin deficiency. I. Usefulness of serum methylmalotilt acid and total homocysteine concentrations. Am J Hematol 1990; 34: 90-8.
246
March 1994
The American Journal of Medicine
--
17. Allen RH, Stabler SP, Savage DG, Lindenbaum J. Elevation of 2-methylcitric acid I and II in the serum, urine and cerebrospinal fluid of patients with cobalamin deficiency. Metabolism 1993; 42: 278-88. 18. Kang SS, Wong PWK, Norusis M. Homocysteinemia due to folate deficiency. Metabolism 1987: 36: 458-62. 19. Chu RC, Hall CA. The toal serum homocysteine as an indicator of vitamin 812 and folate status. Am&J Clin Path 1988; 90: 446-9. 20. Pezzimenti JF, Lindenbaum, J. Megaloblastic anemia associated with erythroid hypoplasia. Am J Med 1972; 53: 748-54. 21. Lindenbaum J, Pezzimenti JF, Shea N. Srall iniestinal function in vitamin B12 deficiency. Ann Intern Med 1974; 80: 326-31. 22. Ballard HS, Lindehbaum J. Megaloblastic anemia complicating hyperalimentation therapy. Am J Med 1974; 56: 740-Z. 23. Lindenbaum J, Roman MR. Nutritional anemia in alcoholism. Am J Clin Nutr 1980; 33: 2727-35. 24. Savage D, Lindenbaum J. Relapses after interrudtion of cyanocobalamin therapy in patients with pernicious anemia. Am J M&d 1983; 74: 765-772. 25. Savage D, Lindenbaum J. Anemia in alcoholics. Medicine 1986; 65: 322-38. 26. Healton EH, Savgge DG, Brust JCM, Garrett TJ, Lindenbaum J. Neurologic aspects of cobalamin deficiency,,yedicine 1991; 70: 229-4s. 27. Spray GH. An improved metHdd for the rapid estimation of vitamin 812 in serum. Clin Sci 1955; 14: 661-7: 28. Herbert V. Minimal daily adult folate requirement. Arch Intern Med 1962; 110: 649-52. 29. Marcell PD, Stabler SP, Pbdell ER, Allen RH. Quantitation of methylmalonic acid and other dic&boxylic acids in normal serum and urine using capillary gas chromatography-mass spectrometty. Anal Biochem 1985; 150: 58-66. 30. Stabler SP, Marcell PD, Podell ER, Allen RH. Quantitation of total homocysteine, total cysteine, and mbthionine in normal serum and urine using capillary gas chromatography-mass spectrometry. Anal Biochem 1987; 162: 185-96. 31. Snedecor (;W, Cochran WG. Statistical methods. Ames, Iowa: Iowa University Press, 1973. 32. Carmel R, Johnson CS. Racial patterns in pernicious anemia: early age at onset and increased frequency of intrinsic-factor antibody in black women. i\j Engl J Med 1978; 298: 647-50. 33. Solanki DL, Jacobson RJ, Green R, McKibbon J, Berdoff R. Pernicious anemia in blacks: a study of 64 patients from Washington, D.C. and Johannesburg, South Africa. Am J Clin Path 1981; 75: 96-9. 34. Carmel R, Johnson CS, Weiner JM. Pernicious anemia in Latin Americans is not a disease of the elderly. Arch Intern Med 1987; 147: 1995-6. 35. Pennypacker LC, Allen RH, Kelly JP, et al. High prevalehce of cobalamin deficiency in elderly outpatients. J Amer Ger Sot 1992; 40: 1197-1204. 36. Hall BE, Watkins CH. Experience with pteroylglutamic (synthetic folic) acid in the treatment of pernicious anemia. J Lab Clin Med 1974; 32: 622-34. 37. Conley CL, Krevans JR. Development of neurologic manifesiations of pernicious anemia during multivitamin thbrapy. N Engl J Med k971; 245: 529-31. 38. Clarke R, Daly L, Robinson K, et al. Hyperhomocysteinemia: an independent risk factor for vascular disease. N Engl J Med 1991; 324: 1149-55. 39. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis Jr. RB, editor. Atherosclerotic cardiovascular disease, hemostasis, and endothelial function. New York, NY: Marcel Dekker, Inc., 1992; 183-236. 40. Stampfer MJ, Malinow MR, Willett WC, et al. A prospective study of plasma homocysteine and risk of myocardial infarction in US physicians. JAMA 1992; 268: 877-81. 41. Eichner ER, Buchanan B, Smith JW, Hillman RS. Variation in the hematologic and medical status of alcoholics. Am J Med Sci 1972; 263: 35-42. 42. Wu A, Chanarin I, Slavin G, Levi AJ. Folate deficiency in the alcoholic-its relationship to clinical and haematological abnormalities, liver disease and folate stores. Br J Haematol 1975; 29: 469-78. 43. Heidemann E, Nerke 0, Waller HD. Alkoholtoxische Veranderungen der Hamatopoiese: Eine prospektive studie bei chronischen alkoholikern. Klin Wochenschr 1981; 59: 1303-12.
Volume 96
March
homocysteine levels were increased in 91% of the 123 episodes of folate deficiency. Methylmalonic acid was elevated in 12.2% of the folate-deficient patients; in all but one, the elevation was attributable to renal insufficiency or hypovolemia. CONCLUSIONS: For the cobalamin-deficient patients, measuring serum metabolite concentrations proved to be a highly sensitive test of deficiency. We conclude that normal levels of both methylmalonic acid and total homocysteine rule out clinically significant cobalamin deficiency with virtual certainty. obalamin (vitamin Blz) deficiency is common. Early diagnosis and treatment can prevent severe anemia or potentially irreversible damage to the nervous system from developing, but the majority of patients lack the textbook features of advanced de& ciency of the vitamin [l-4]. Serum cobalamin concentration is widely used as a screening test for deficiency but has major limitations. Its sensitivity is less than perfect [5,6], and many workers [4,7-121 have found that a substantial proportion-perhaps 25% to 50% [4,12]of patients with low serum cobalamin levels do not appear to be deficient in the vitamin or to have underlying disorders predisposing them to deficiency. The metabolites methylmalonic acid and homocysteine accumulate when the two mammalian cobalamin-dependent enzymatic reactions are impaired [ 13,141. We found that serum levels of both metabolites are markedly elevated (more than three standard deviations [SD] above the mean in healthy control subjects) in most deficient patients, suggesting that these tests are a useful and convenient way to establish a diagnosis of cobalamin deficiency [4&L-15]. The assays were modified in early 1988 to further increase their specificity and precision [ 16,171. We report here using the modified techniques to assay more than 400 patients with clinical evidence of cobalamin deficiency. To our knowledge, this series, accumulated over 21 years, is the largest group of patients with well-documented cobalamin deficiency reported in the literature. For comparison, we assayed a series of 119 patients with folate deficiency. In previous studies involving relatively small numbers of folate-deficient patients, total homocysteine concentrations were elevated in most cases [14,18,19].
C
1994
The American
Journal
of Medicine
Volume
96
239
TABLE I Criteria Serum
TABLE III
for Cobalamin Cobalamin
Deficiency
< 200 pg/mL
Diagnostic Marrow +/or Blood Smear +
in 434 Episodes
in 401 Patients
Criteria
Response to Cbl Rx*
No. of Episodes 309 13 107 5
N+A + +
rTA 0 Total
% 71.2 3.0 24.6 1.2
434
100
NA = not available; Cbl = cobalamin; Rx = therapy. *See Table II for definition of responses.
TABLE II Responses Deficiency
to Cobalamin Treatment in 409 Patients
in 434 Episodes
Decrease in mean cell volume > 5 fL Increase in hematocrit 2 0.05 Resolution of thrombocytopenia Resolution of leukopenia Improvement in neuropsychiatric abnormalities* More than one of above responses to treatment Other responsest No documentation of response
of Cobalamin
No. of Episodes 369 286 110 50 112 325 9 13
.g 85.0 65.9 25.3 11.5 25.8 74.9 2.1 3.0
*Clear-cut improvement in neuropsychiatric abnormalities within the first three months of treatment. This was the only response to treatment in a single episode. TOnly one of the following responses was documented: reticulocytosis (five patients); correction of neutrophil hypersegmentation (three patients); and resolution of atrophic glossitis (one patient).
PATIENTS AND METHODS Patients All cases of cobalamin and folate deficiency seen at Harlem Hospital Center and Columbia-Presbyterian Medical Center between July 1968 and June 1989 were reviewed. The majority of the patients were evaluated in consultation with one of the coauthors as part of prospective studies of megaloblastic anemias, and the hematologic and neurologic findings have been the subject of previous reports [4,13-15,20-261. Methylmalonic acid and total homocysteine levels were measured in all patients seen between 1968 and 1981 who met the study’s criteria for cobalamin and folate deficiency and from whom serum stored at -20°C was available. From 1982 to 1989, serum metabolite levels were measured in each of 267 consecutively seen patients who met the cirteria for cobalamin deficiency (219 patients) or folate deficiency (48 patients). Methods When available, smears of peripheral blood and bone marrow from cobalamin- and folate-deficient patients were coded and interspersed with those of patients without evidence of deficiency for blinded review by one of the authors. Hypersegmentation in 240
March
1994
in 119 Patients No. of Episodes Diagnostic marrow or blood smear, or both 123 Serum folate i 4 ng/mL* 123 Serum cobalamin > 300 pg/mLT 123 Underlying disorder associated with folate deficiency 123 Documented response to folic acid 94
and
The American
Journal
of Medicine
Volume
for Folate Deficiency
*Serum folate < 2.1 ng/mL TSerum cobalamin 301 to 10, 401 to 500 pg/mL in 1,000 pg/mL in 26. None before study.
in 123 Episodes
?j 100 100 100 100 76.4
in 99 episodes and 2.1 to 3.9 ng/mL in 24. 359 pg/mL in 7 episodes, 351 to 400 pg/mL in 20, 501 to 1,000 pg/mL in 60, and greater than of the patients had been treated with vitamin EP*
blood smears was defined as more than five cells with five lobes per 100 neutrophils or any cells with six or more lobes. Serum cobalamin levels were determined by radioassay using purified intrinsic factor (Quantaphase, Bio-Rad Laboratories, Richmond, CA) or by microbiologic assay with LactobaciUus leichmannii [27]. Serum folate concentrations were measured by milk-binder radioassays or by microbiologic assay with Lactobacillus casei [28]. Serum concentrations of methylmalonic acid 1291 and total homocysteine [30] were measured by modified [ 16,171 techniques using capillary-gas chromatography and mass spectrometry. The normal ranges (calculated as the mean & 3 SD after log transformation to correct for skewing toward higher values) are 53 to 376 nmol/L for methylmalonic acid and 4.1 to 21.3 urnol/L for total homocysteine [ 161. For 160 of the patients with cobalamin deficiency and 19 with folate deficiency, serum metabolite levels were measured by methods used before our modifications were made, and the results were reported earlier [4,13-151. Standard statistical methods, including Student’s t-test and the &i-squared test with Yates’ correction for continuity [31], were used to analyze the data
RESULTS Criteria of Deficiency Serum methylmalonic acid and total homocysteine levels were measured in 406 patients who had 434 episodes of cobalamin deficiency; 25 patients had one or more recurrent episodes of deficiency because they discontinued maintenance treatment with the vitamin. The diagnostic criteria for including cobalamindeficient patients in this series are summarized in Table I. All patients had low serum cobalamin levels. Morphologic abnormalities (megaloblastic bone marrow or hypersegmented neutrophils, or both, with or without macroovalocytes, on blood smear) were found in 322 episodes; clear-cut responses to cobalamin therapy (as defined in Table II) were documented for 309 of the 322 episodes. Although marrow and blood films were not available for 107 episodes before therapy began, clear-cut responses were doc96
SERUM METABOLITES IN COBALAMIN AND FOLATE DEFICIENCY / SAVAGE ET AL TABLE IV Etiology
of Cobalamin Etiology
TABLE V and Folate
Deficiencies
in 529 Patients Patients
Cobalamin deficiency Pernicious anemia Proven Probable* Tropical sprue Proven Probablet Gastrectomy lleal disease or resection, or both Gastric and ileal resections Jejunal diverticula Vegetarianism Food cobalamin malabsorption Not established Total
234 76 11 28 13 13 1 6 2 3 19
‘One or more of the following: cobalamin malabsorption Schilling test, achlorhydria, and autoimmune thyroid disease.
37.2 62.8
70/l 19 49/119
58.8 41.2
Ethnicity Black White Latin0 Other
168/406 135/406 101/406 2/406
41.3 33.3 24.9 0.5
105/119 9/l 19 5/119 o/119
88.2 7.6 4.2 0.0
59/434
13.6
107/123
87.0
Anemia
X3/434
72.1
123/123
100.0
Macrocytosis MCV > 100 fL MCV> IlOfL
334/402 224/402
83.1 55.7
68/91 50/91
74.7 54.9
73/433
16.9
22/123
17.9
136/384
35.4
78/l 18
66.1
Pancytopenia
48/384
12.5
21/118
17.8
LDH > 1,000 U/L
96/330
29.1
24/83
28.9
11 l/434
25.6
27/123
22.0
156/434
35.9
36/123
29.3
ThrombocytopeniaS
Atrophic glossitis Neuropsychiatric
in part 1 of the
in a person who had lived in the tropics
151/406 255/406
Leukopeniat
119
(intestinal
unrented; all but one of these episodes showed at least one of the hematologic responses listed in Table II. In five episodes showing a therapeutic response, pretreatment blood smears were normal or nondiagnostic (marrow aspirates were not obtained). Of the 434 episodes, 421 (97%) responses to treatment with cyanocobalamin were documented (Table ID. Serum metabolites were measured in 123 episodes of folate deficiency diagnosed in 119 patients (four patients had two episodes). Our definition of folate deficiency was based on the clinical, morphologic, and biochemical criteria summarized in Table III. Hematologic responses to treatment with folic acid were documented for 94 episodes; adequate hematologic follow-up data were not obtained for the others. Patients were considered folate or cobalamin deficient based solely on the stated criteria and not on the results of the metabolite assays. The majority of the patients (86%) were studied before 1986, when the metabolite assays became available, and the percent of patients with each deficiency with elevated metabloite levels was similar for patients who were evaluated before and after 1986.
CAUSES OF DEFICIENCY STATES Diagnostic studies were adequate to determine the etiology of cobalamin deficiency for most patients March
Folate Deficiency No./Totalt %
Sex Male Female
Alcoholism
103 11 1 1 1 1 1
Total
Cobalamin Deficiency No./TotalT %
Characteristics*
406
Folic acid deficiency Alcoholism Malnutrition Malabsorption syndrome ? Oral contraceptives Pregnancy Sickle cell anemia Tropical sprue
TMalabsorption syndrome biopsy not performed).
Clinical Characteristics of 434 Episodes of Cobalamin Deficiency in 406 Patients and 123 Episodes of Folate Deficiency in 119 Patients
syndromes
MCV = mean corpuscular volume; LDH = lactate dehydrogenase. *Mean age 49 * 17 years in folate-deficient group and 66 f 18 years in cobalamin-deficient patients. tTotal number of cases for which information was available; denominator = number of patients, for sex and ethnic background, and number of episodes for other items. SLeukopenia = leukocyte count less than 4,OOO/pL for white and Latin0 patients and less than 3,OOO/pL for blacks; thrombocytopenia = platelet count less than 150,000/$.. .$Consistent with those syndromes seen in cobalamin deficiency f261. Peripheral neuropathy, ataxia, and cerebral dysfunction were often seen in folate-deficient alcoholics.
(Table IV). Pernicious anemia, demonstrated by serial Schilling tests or serum antibody to intrinsic factor, was established for 57.6% and considered likely for another 18.7%. The serum folate level was less than 2.1 ng/mL in 17 patients and between 2.1 ng/mL and 3.9 ng/mL in 57 patients in the cobalamindeficiency group. All 74 patients had one or more fmdings of serum antibodies to intrinsic factor, malabsorption of vitamin BLzin Schilling tests, or neurologic dysfunction responsive to cobalamin treatment. Of these 74 patients, 19 had proven or probable tropical sprue, and it is likely that some had folate deficiency along with cobalamin deficiency. The 19 cobalamindeficient patients for whom diagnostic evaluation was inadequate to determine the etiology of the deficiency (Table IV) had serum folate levels greater than 4 ng/mL in all cases. Alcoholism caused 87% of the episodes of folate deficiency (Table IV).
Characteristics of Episodes The clinical features of the cobalamin- and folatedeficient patients are summarized in Table V. The 1994
The American
Journal
of Medicine
Volume
96
241
TABLE VI Serum Concentrations of Methylmalonic Homocysteine in Episodes of Cobalamin Without Anemia
Acid Deficiency
and Total With and
AnemiaPresent*AnemiaAbsent All Episodes No. % No. % No. % Total numberof episodes 313 121 434 Episodeswith ElevatedsMMA 308 98.4 119 98.3 427 98.4 ElevatedsHCYS 306 97.8t 110 90.9t 416 95.9 Both MMA, HCYS elevated MMA alone elevated HCYS alone elevated MMA, HCYS both normal
302 6 4 1
96.5t 1.9 1.3 0.3
108 11 2 0
89.3$ 9.2 1.7 0
410 17 6 1
94.5 3.9 1.4 0.2
MMA= methylmalonicacid; HCYS= total homocysteine.
. 50
100
150
200
250
*Hematocrit corpuscular anemia was tp <0.005. $p
300 ‘W%5
HOMOCYSTEINE, pmol/L
.
less than 40% in males; less than 35% in females. The mean volume was normal (80 to 100 fL) in 9.9% of the episodes when present and in 33.3% of those when anemia was absent.
3 SD above the mean in normal control subjects.
TABLE VII
.
I
Serum Metabolite Concentrations in Anemia Caused by Cobalamin or Folate Deficiency Cobalamin Deficiency Folate Deficiency No. 96 No. %
.
Total no. of episodes*
297
Elevated MMA
292
98.3
98
Elevated HCYS
290
97.6
88
89.8
Both MMA,HCYSelevated
4.1
4t
4.1
286
96.2
4
Elevated MMA, normal HCYS
6
2.0
0
Elevated HCYS. normal MMA
4
1.3
84
85.7
MMA,HCYSboth normal
1
0.3
10
10.2
0
MMA= methylmalonicacid; HCYS= total homocysteine. *Sixteen patients with cobalamin deficiency and 25 with folate deficiency elevated serum creatinine levels were excluded from the analysis.
with
t.Severe volume depletion present in three episodes. .-
a--,--------------
I
50
100 HOMOCYSTEINE,
L
150 FmoliL
I
200 220
Figure 1 Top. Serum concentrations of methylmalonic acid and total homocysteine in 313 episodes of megaloblastic anemia due to cobalamin deficiency. Dashed lines indicate three standard deviations above the mean for normal controls for each metabolite; hatched area includes values for both metabolites less than three standard deviations above the mean for normal controls; open circles indicate 16 cobalamin deficiency episodes associated with anemia with elevated serum creatinine. Figure 1 Bottom. Serum cdncentrations of methylmalonic acid and total homocysteine in 121 episodes of cobalamin deficiency in which the hematocrit was normal (greater than or equal to 40% in men, greater than or equal to 35% in women). Dashed lines indicate three standard deviations above the mean for normal controls for each metabolite; hatched area includes values for both metabolites less than three standard deviations above the mean for normal controls. 242
March 1994
The American Journal of Medicine
substantial numbers of black and Latin0 patients with cobalamin deficiency reflect the distribution of ethnic groups in the catchment areas of the two hospitals and are consistent with previous reports of a high prevalence of pernicious anemia in these populations [32-34]. The higher proportion of black patients and anemic patients in the folate-deficient group came from including several series of alcoholics with anemia studied at Harlem Hospital Center [23,25]. The mean corpuscular volume (MCV) was normal in 17% of cobaknnin-deficient patients tid in 25% of folate-deficient patients. Only a minority of patients with either deficiency state had leukopenia, pancytopenia, atrophic glossitis, or marked elevations of serum lactate dehydrogenase. More than a third of the cobalamindeficient group has neuropsychiatric symptoms and signs (Table V) compatible with that vitamin deficiency [26]. Peripheral neuropathy, ataxia, and cerebral dysfunction were frequently associated with alcoholism in the folatedeficient patients.
Volume 96
SERUM
METABOLITES
Serum Metabolites in Cobalamin Deficiency Serum methylmalonic acid levels were markedly increased (more than 3 SD above the mean in normal controls) in 427 (98.4%) of the 434 episodes of cobalamin deficiency (Figure 1, Table VI); serum total homocysteine concentrations were markedly increased (more than 3 SD above the mean in normal controls) in 416 (95.9%). In 410 episodes (94.50/o),both metabolite levels were markedly elevated. In only one cobalamin-deficient patient (0.2%) were both metabolites normal. The hematocrit was normal (Figure 1 Bottom; Table VI) in 121 of the 434 episodes of cobalamin deficiency. Nonetheless, serum methylmalonic acid values were elevated in 98.3% of the episodes without anemia In contrast, homocysteine levels were elevated in 90.9% of the nonanemic patients compared with 97.8% of anemic patients (Table VI, chi squared = 8.66, p ~0.005). Thus serum methyhnalonic acid levels were more sensitive for identifying nonanemic cobalamindeficient patients than serum homocysteine levels (p ~0.005). Both metabolites were elevated in 89.3% of the episodes in which the hematocrit was normal. Mean homocysteine levels were significantly higher in anemic than nonanemic patients (89.4 + 55.0 umol/L versus 60.2 t- 41.3 pmol/L, p
IN COBALAMIN
IO
20
AND
50
FOLATE
DEFICIENCY
100
150
HOMOCYSTEINE,
/ SAVAGE
200
ET AL
250
pmol/L
Figure 2. Serum concentrations of methylmalonic acid and total homocysteine in 123 episodes of megaloblastic anemia due to folate deficiency. Open circles indicate patients with elevated serum creatinine concentrations; boxes indicate three patients with severe hypovolemia and normal creatinine levels. Dashed lines indicate three standard deviations above the mean for normal controls. (Note that the scale for methylmalonic acid values differs from that in Figure 1.)
obtained during the most recent 5 years of the study, 1984 to 1989 (data not shown). In a series of 86 consecutively studied patients with cobalamin deficiency [4] tested by earlier assay methods, 12 (14%) did not have elevated methylmalonic acid and 13 (15.1%) did not have increased homocysteine. Sera were available from most of these patients for assay by the modified techniques used in this study. Seven of 10 patients with methylmalonic acid levels less than 3 SD above the normal mean using the older method had elevated concentrations using the current assay system. Of 11 previously normal homocysteine values, four were increased when reassayed by the modified method. Serum Metabolites in Folate Deficiency Serum homocysteine levels were markedly elevated in 112 (91.0%) of the 123 episodes of folate deficiency (Figure 2). The 11 patients with normal levels did not differ from the others in severity of anemia, megaloblastic morphologic changes, methylmalonic acid elevation, time in hospital before study, reticulocyte counts, or serum levels of folate, cobalamin, creatinine, bilirubin, or albumin (data not shown). Serum methylmalonic acid was increased in 15 (12.2%) of the 123 episodes of folate deficiency (Figure 2); however, 14 of these patients had renal dysfunction (11 with serum creatinine elevation and 3 with clinical evidence of severe volume depletion). 1994
The American
Journal
of Medicine
Volume
96
243
Excluding these 14 patients leaves serum methylmalonic acid elevation in only one episode of folate deficiency; and in that patient, the increase was modest (398 &I/L). Serum Metabolites in Cobalamin Versus Folate Deficiency The serum metabolite levels of the folate-deficient patients, all of whom were anemic, were compared with those of the cobalamin-deficient patients with anemia (Table VII). Patients with elevated serum creatinine values (16 with cobalamm deficiency and 25 with folate deficiency) were excluded from the analysis. Serum methylmalonic acid was markedly elevated in 98.3% of the 297 cobalamin deficiency episodes and in only 4.1% of the 98 folate deficiency episodes (chi squared = 343, p
March
1994
The American
Journal
of Medicine
Volume
COMMENTS Concentrations of methylmalonic acid and homocysteine are stable indefinitely in serum stored at -20°C [13,14], and we found no trend toward lower values in serum that had been repeatedly thawed and frozen over a period as long as 20 years. Hence, the stability of these compounds allowed us to test their diagnostic usefulness for a large series of patients with well-documented cobalamin and folate deficiency that would have required decades to assemble prospectively. We included all patients with these deficiencies seen at two New York City hospitals between 1968 and 1989 from whom serum was available, after excluding patients with low or borderline levels of both vitamins when it was uncertain which deficiency they had. The results point to the high sensitivity of serum methylmalonic acid and homocysteine concentrations, using the modified assay methods, for diagnosing patients with clinically significant cobalamin deficiency. One or the other metabolite level was markedly elevated (greater than 3 SD above the mean in normal controls) in all but one of the 434 episodes (Table VI, Figure l), giving a combined sensitivity of 99.8% for the metabolites for detecting deficiency of the vitamin. While methylmalonic acid and homocysteine concentrations were both increased in approximately 98% of cobalamin-deficient patients with megaloblastic anemia, the methylmalonic acid level was more consistently elevated in nonanemic patients (Table VI). Using the modified techniques, methylmalonic acid determinations also had greater sensitivity for detecting mild or early cobalamin deficiency in patients with pernicious anemia receiving infrequent maintenance therapy with cobalamin [6] as well as in elderly patients with low-normal serum cobalamin levels [35]. Moelby and colleagues also found serum methylmalonic acid elevated in 30 of 31 patients with cobalamin deficiency [ 121. In spite of the generally high sensitivity of serum methylmalonic acid determinations, concentrations were normal in 7 of 434 episodes of cobalamin deficiency (Table VI); however, the homocysteine level was increased in six of the seven episodes, indicating that the two metabolite determinations combined improve the rate of detecting cobalamin deficiency over that of methylmalonic acid alone. The reason patients occasionally fail to show a methylmalonic acid elevation is not clear, although gut flora alterations that disturb propionate production by enteric bacteria may play a role [61. We have included in the cobalamin-deficient group 74 patients whose serum folate levels were also low or at the lower end of the normal range. Most of these patients had underlying pernicious anemia, documented by serum antibodies to intrinsic factor, or 96
SERUM METABOLITES IN COBALAMIN AND FOLATE DEFICIENCY / SAVAGE ET AL
cobalamin malabsorption corrected by intrinsic factor in Shilling tests performed many weeks after correcting the deficiency state. In some of the patients with tropical sprue, however, folate deficiency may have coexisted with cobalamin deficiency, or conceivably, may have been the primary deficiency. Serum methylmalonic acid was elevated in 72 and serum homocysteine in all of the 74 patients considered cobalamin deficient who had low or low-normal folate values. Excluding these patients from the analysis would not significantly change any of the percentages listed in Table K-in fact, the percentage of patients with cobalamin deficiency and increased methylmalonic acid level would be even higher. Patients with folate deficiency would be expected to have a block in the conversion of homocysteine to methionine [14]; and indeed, serum homocysteine was markedly increased in 91% of the 123 episodes of folate deficiency, confiig earlier reports of smaller numbers of patients [ 14,18,19]. The pattern of elevated homocysteine associated with normal methylmalonic acid in a patient with a megaloblastic anemia was much more commonly associated with folate deficiency than with cobalamin deficiency: 84 of 88, or 95% of such episodes (Table VII) were caused by folate deficiency. Serum methylmalonic acid would not be expected to accumulate as a result of lack of folate; however, it was increased in 15 (12%) of the 123 episodes of folate deficiency. Methylmalonic acid is excreted in the urine. Eleven of the 15 patients with folate deficiency and elevated serum methylmalonic acid had renal function impairment severe enough to increase serum creatinine; three other patients had marked intravascular volume depletion, which may have led to renal retention of methylmalonic acid; for only one folatedeficient patient was there no explanation for the elevation. Therefore, impaired renal function appears to be the principal cause of a false-positive increase in serum methylmalonic acid in patients who do not appear to be cobalamin deficient. For the majority of patients undergoing evaluation for possible cobalamin deficiency, available clinical and laboratory data on renal function and intravascular volume status will indicate whether changes in these conditions may be responsible for an elevated methylmalonic acid. Furthermore, renal dysfunction by itself appears to cause only a modest increment in the serum methylmalonic acid level (Figure 2); for most patients with cobalamin deficiency, the methylmalonic acid concentration exceeds that attributable to renal insufficiency alone (compare Figures 1 and 2). We are currently studying a large series of patients with renal disorders and no cobalamin or folate deficiency to determine the upper limit of the methylmalonic acid level caused by disturbances in kidney function. March
Although total homocysteine is elevated in both cobalamin and folate deficiency, making the correct diagnosis is important since treating a cobalamindeficient patient with folic acid alone will not prevent or correct the neuropsychiatric abnormalities caused by the deficiency, even though partial or even complete hematologic responses to folic acid often occur 136,371. In addition, elevated homocysteine will respond only to treatment with the proper vitamin this is true even when cobalamin-deficient patients have hematologic responses to folic acid [ 161. Normalizing an elevated homocysteine level may be important in itself since a number of recent studies have shown that elevated levels are strongly correlated with coronary artery, cerebrovascular, and peripheral vascular disease [38-40]. Some alcoholic patients with serum folate concentrations between 2.1 and 4.0 ng/mL (i.e., at the lower end of the normal range) were included in the folatedeficient group: all had megaloblastic anemia, unequivocally normal serum cobalamin values (Table III), and clear-cut responses to folic acid treatment, and all were considered highly likely to be folate deficient. A number of previous studies of alcoholic patients have noted normal serum folate values (commonly in the low-normal range) in 25% to 55% of patients with florid evidence of folate deficiency [25,4143]. Our findings from this large series of patients indicate that serum metabolite levels are highly sensitive tests for diagnosing cobalamin deficiency-normal levels of both methylmalonic acid and homocysteme rule out a deficiency with virtual certainty (Table VI). Furthermore, an elevated serum level of methylmalonic acid (or of both methylmalonic acid and homocysteine) is much more specific for cobalamin deficiency than is a low serum cobalamin level, which is frequently seen without any other evidence of deficiency [4,7-121. The only important limitation to the specificity of an increased serum methylmalonic acid concentration is underlying renal dysfunction, which the great majority of cobalamin-deficient patients do not have. Inborn errors of metabolism may lead to elevated serum homocysteine or methylmalonic acid; based on our experience, other causes of falsepositive serum metabolite elevations appear to be quite rare. Measurements of serum levels of these metabolites, available through commercial laboratories, appear to be highly useful ancillary tests for interpreting low or low-normal serum cobalamin concentrations, particularly since many, if not most, deficient patients lack the full-blown textbook manifestations of megaloblastic anemia (Table V) [14]. The availability of metabolite testing may obviate the need for Schilling tests or for prolonged therapeutic trials that may yield 1994
The American
Journal
of Medicine
Volume
96
246
SERUM METABOLITES IN COBALAMIN AND FOLATE DEFICIENCY / sAVAGE ET AL l__l~-.._.
equivocal results for many patients. Although the Schilling test measures cobalamin absowtion rather than deficiency, it is often used as a secondary test when a diagnosis of cobalamin deficiency is uncertain, on the assumption that deficiency is more likely if the Schilling test is abnormal. The metabolites provide more direct evidence of deficiency of the vitamin. Further studies will be required to determine whether metabolite level measurements should serve mainly as backup tests after serum cobalamin and folate values are obtained for patients with suspected deficiencies, or whether they should replace serum vitamin concentrations as the prim&y screening tests for these disorders.
REFERENCES 1. Blundell EL, Matthews JH, Allen SM, Middleton AM, Morris JE, Wickramasinghe SN. Importance of low serum vitamin B12 and r@d cell folate concentrations in elderly hospital inpatients. J Clin Path 1985; 38: 1179-1184. 2. Carmel R. Pernicious anemia-the expected findings of very low serum cobalamin levels, anemia, and macrocytosis are often lacking. Arch Intern Med 1988; 148: 1712-4. 3. Matthews JH, Clark DM, Abrahamson GM. Effect of therapy with vitamin Bli and folic acid on elderly patients with low concentrations of serum vitamin 812 or erythrocyte folate but normal blood counts. Acta Haemat 1988; 79: 84-87. 4. Stabler SP, Allen RH, Savage DG, Lindenbaum J. Clinical spectrum ‘and diagnosis of cobalamin deficiency. Blood 1990; 76: 871-81. 5. Magnus FM. Cobalamin and unsaturated transcobalamin values in pernicious anaemia: relation to treatment. Stand J Haematol 1986; 36: 457-65. 6. Lindenbaum J, Savage D, Stabler SP, Allen RH. Diagnosis of cobalamin deficiency. II. Relative sensltlvrtles of serum cobalamin, methylmalonic acid and total homocysteine concentrations. Am J Hematol 1990; 34: 99-107. 7. Bain B, Broom GN, Woodside J, Litwinczuk RA, Wickramasinghe SN. Assessment of a radioisotopic assay for vitamin B12 using Bn intrinsic factor preparation with R proteins blocked by vitamin B12 analogues. J Clin Path 1982; 35: 1110-3. 8. Schilling RF, Fairbanks VF, Miller R, Schmitt K, Smith MJ. Improved vitamin B12 assays, a report on two commercial kits. Clin Chem 1983; 29: 582-3. 9. Brynskov J, Andersen K, Gimsing P, Hippe E. False low serum vitamin B12 values with radiodilution assays using blocked R-binders. Lancet 1983; 1: 1104-5. : 10. Cooper BA, Fehedy V, Blanshay P. Recognition of deficiency of vitamin Bi2 using measurement of serum concentration. J Lab Clin Med 1986; 107: ‘: 447-52. / 11. Thompson WG, Cassino C, Babitz L, et al. Hypersegmented neutrophils !and vitamin 812 deficiency. Acta Haemat 1989; 81: 186-191. ‘;12. Moelby L, Rasmussen K, Jensen MK, Pedersen KO. The relationship between clinically confirmed cobalamin deficiency and serum methylmalonic iacid. J Int Med 1990; 228: 373-378. ‘13. Stabler SP, Marcell PD, Podell ER, Allen RH, Lindenbaum J. Assay of methylmalonic acid in the serum of patients with cobalamin deficiency using capillary gas chromatography-mass spectrometry. J Clin Invest 1986; 77: 1606-12. 14. Stabler SP, Marcell PD, Podell ER, Allen RH, Savage D, Lindenbaum J. Elevation of total homocysteine in the serum of patients with cobalamin or folate deficiency detected by capillary gas chromatography-mass spectrometry. J Clin Invest 1988; 81: 466-74. 15. Lindenbaum J, Healton EB, Savage D, et al. Frequency of neuropsychiatric disorders caused by cobalamin deficiency in the absence of anemia or marcocytosis. New Engl J Med 1988; 318: 1720-8. 16. Allen RH, Stabler SP, Savage D, Lindenbaum J. Diagnosis of cobalamin deficiency. I. Usefulness of serum methylmalotilt acid and total homocysteine concentrations. Am J Hematol 1990; 34: 90-8.
246
March 1994
The American Journal of Medicine
--
17. Allen RH, Stabler SP, Savage DG, Lindenbaum J. Elevation of 2-methylcitric acid I and II in the serum, urine and cerebrospinal fluid of patients with cobalamin deficiency. Metabolism 1993; 42: 278-88. 18. Kang SS, Wong PWK, Norusis M. Homocysteinemia due to folate deficiency. Metabolism 1987: 36: 458-62. 19. Chu RC, Hall CA. The toal serum homocysteine as an indicator of vitamin 812 and folate status. Am&J Clin Path 1988; 90: 446-9. 20. Pezzimenti JF, Lindenbaum, J. Megaloblastic anemia associated with erythroid hypoplasia. Am J Med 1972; 53: 748-54. 21. Lindenbaum J, Pezzimenti JF, Shea N. Srall iniestinal function in vitamin B12 deficiency. Ann Intern Med 1974; 80: 326-31. 22. Ballard HS, Lindehbaum J. Megaloblastic anemia complicating hyperalimentation therapy. Am J Med 1974; 56: 740-Z. 23. Lindenbaum J, Roman MR. Nutritional anemia in alcoholism. Am J Clin Nutr 1980; 33: 2727-35. 24. Savage D, Lindenbaum J. Relapses after interrudtion of cyanocobalamin therapy in patients with pernicious anemia. Am J M&d 1983; 74: 765-772. 25. Savage D, Lindenbaum J. Anemia in alcoholics. Medicine 1986; 65: 322-38. 26. Healton EH, Savgge DG, Brust JCM, Garrett TJ, Lindenbaum J. Neurologic aspects of cobalamin deficiency,,yedicine 1991; 70: 229-4s. 27. Spray GH. An improved metHdd for the rapid estimation of vitamin 812 in serum. Clin Sci 1955; 14: 661-7: 28. Herbert V. Minimal daily adult folate requirement. Arch Intern Med 1962; 110: 649-52. 29. Marcell PD, Stabler SP, Pbdell ER, Allen RH. Quantitation of methylmalonic acid and other dic&boxylic acids in normal serum and urine using capillary gas chromatography-mass spectrometty. Anal Biochem 1985; 150: 58-66. 30. Stabler SP, Marcell PD, Podell ER, Allen RH. Quantitation of total homocysteine, total cysteine, and mbthionine in normal serum and urine using capillary gas chromatography-mass spectrometry. Anal Biochem 1987; 162: 185-96. 31. Snedecor (;W, Cochran WG. Statistical methods. Ames, Iowa: Iowa University Press, 1973. 32. Carmel R, Johnson CS. Racial patterns in pernicious anemia: early age at onset and increased frequency of intrinsic-factor antibody in black women. i\j Engl J Med 1978; 298: 647-50. 33. Solanki DL, Jacobson RJ, Green R, McKibbon J, Berdoff R. Pernicious anemia in blacks: a study of 64 patients from Washington, D.C. and Johannesburg, South Africa. Am J Clin Path 1981; 75: 96-9. 34. Carmel R, Johnson CS, Weiner JM. Pernicious anemia in Latin Americans is not a disease of the elderly. Arch Intern Med 1987; 147: 1995-6. 35. Pennypacker LC, Allen RH, Kelly JP, et al. High prevalehce of cobalamin deficiency in elderly outpatients. J Amer Ger Sot 1992; 40: 1197-1204. 36. Hall BE, Watkins CH. Experience with pteroylglutamic (synthetic folic) acid in the treatment of pernicious anemia. J Lab Clin Med 1974; 32: 622-34. 37. Conley CL, Krevans JR. Development of neurologic manifesiations of pernicious anemia during multivitamin thbrapy. N Engl J Med k971; 245: 529-31. 38. Clarke R, Daly L, Robinson K, et al. Hyperhomocysteinemia: an independent risk factor for vascular disease. N Engl J Med 1991; 324: 1149-55. 39. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and cardiovascular disease. In: Francis Jr. RB, editor. Atherosclerotic cardiovascular disease, hemostasis, and endothelial function. New York, NY: Marcel Dekker, Inc., 1992; 183-236. 40. Stampfer MJ, Malinow MR, Willett WC, et al. A prospective study of plasma homocysteine and risk of myocardial infarction in US physicians. JAMA 1992; 268: 877-81. 41. Eichner ER, Buchanan B, Smith JW, Hillman RS. Variation in the hematologic and medical status of alcoholics. Am J Med Sci 1972; 263: 35-42. 42. Wu A, Chanarin I, Slavin G, Levi AJ. Folate deficiency in the alcoholic-its relationship to clinical and haematological abnormalities, liver disease and folate stores. Br J Haematol 1975; 29: 469-78. 43. Heidemann E, Nerke 0, Waller HD. Alkoholtoxische Veranderungen der Hamatopoiese: Eine prospektive studie bei chronischen alkoholikern. Klin Wochenschr 1981; 59: 1303-12.
Volume 96