Trans-blood-brain-barrier albumin leakage and comparisons of intrathecal IgG synthesis calculations in multiple sclerosis patients

Journal of Neuroimmunology, 46 (1993) 185-192 Elsevier Science Publishers B.V.

185

JNI 02426

Trans-blood-brain-barrier albumin leakage and comparisons of intrathecal IgG synthesis calculations in multiple sclerosis patients K. Syndulko a, W.W. Tourtellotte a, A.J. Conrad ~, G. Izquierdo b, Multiple Sclerosis Study G r o u p *, Alpha Interferon Study G r o u p * and Azathioprine Study G r o u p * Neurology and Research Services, VAMC W. Los Angeles, Los Angeles, CA and Department of Neurology, UCLA School of Medicine, Los Angeles, CA, USA and b Servico de Neurologia, Hospital Universitario, Sevilla, Spain (Received 9 March 1993) (Revision received 1 April 1993) (Accepted 1 April 1993)

Key words." Cerebrospinal fluid; Immunoglobulin G; Albumin; Multiple sclerosis; Intrathecal

Summary We compared four equations for estimating intrathecal IgG synthesis (Tibbling and Link IgG index (T/L), Schuller and Sagar (S/S), Reiber and Felgenhauer (R/F), and Tourtellotte (T) equations) using data from chronic progressive MS patients. For normal albumin leakage (AL) ( < 75 mg/day - intact BBB), T (r = 0.15) and R / F (r = 0.10) showed comparable positive correlations with trans-BBB AL, T / L (r = -0.10) was negative and S / S was uncorrelated (r = 0.05). For abnormal AL (>_ 75 mg/day), the R / F ( r = -0.24), S / S (r = -0.37) and the T / L ( r = -0.22) equations overcorrected, whereas the T (r = 0.07) equation values did not correlate with AL. The albumin index and trans-BBB albumin leakage rate formulae gave essentially identical estimates of BBB leakage (r = 0.99, P = 0.0001). We conclude that in chronic progressive MS patients the R / F , T / L and S / S formulae overcompensate for large abnormal T-BBB albumin leakage rates. The T formula corrected best for IgG transudate at high AL rate values in MS.

Introduction One of the most recognized features of multiple sclerosis (MS) has been the quantitative and qualitative alteration of immunoglobulin G (IgG) in the cerebrospinal fluid (CSF) of patients (Tourtellotte and Walsh, 1984). IgG is synthesized by only one cell type in the body, namely, the plasma cell (Myers, 1991). In health there are no plasma cells in the brain (Prineas, 1985). Normal IgG concentration in the CSF is determined by the concentration in the blood (Reiber and Felgenhauer, 1987; Tourtellotte et al., 1985a) and by the natural permeability of endothelial capillary tight junctions primarily located in the choroid plexus (Brightman, 1965). When infection occurs in the brain,

Correspondence to: W.W. Tourtellotte, M.D., Neurology Service (W127A), VAMC West Los Angeles, Los Angeles, CA 90073, USA. * See Acknowledgements for listing of authors.

B lymphocytes are recruited into the CNS and mature there to plasma cells. The newly recruited plasma cells synthesize IgG. An abnormally elevated IgG concentration in the CSF can also be the result of an elevated blood IgG concentration in conjunction with breaks in the endothelial tight junctions of the blood-brain barrier (BBB). If a correction of CSF IgG concentration for high serum IgG concentration and leakage of IgG across the BBB yields an excess of CSF IgG, then there exists intrathecal IgG synthesis (Tourtellotte et al., 1980). Accordingly, intrathecal IgG synthesis calculations provide a measure of the presence and quantity of plasma cells secreting IgG inside the BBB, i.e., evidence of inflammation in the CNS. Certain measurements have been proposed in order to distinguish between IgG synthesized within the central nervous system (CNS) compartment and IgG transudate from serum. There is a general consensus that intrathecal IgG synthesis occurs in the majority if not all clinical definite MS patients. However, the question

186 of which intrathecal IgG synthesis measurement most accurately reflects the real world has not yet been resolved (Whitaker, 1985). Different formulas have been developed to quantify IgG synthesis in the CNS, and each has its own assumptions and constraints. Tibbling and Link ( T / L ) (Tibbling et al., 1977) created a dimensionless index based on the CSF to serum proportions of IgG and albumin. Quantification of intrathecal IgG synthesis is not possible with this approach (Whitaker, 1985). The Tourtellotte formula (T) (Tourtellotte et al., 1980) is a quantitative measurement of IgG synthesis rate inside the BBB. The formula is based on the proportional crossing of serum IgG and albumin across the BBB. The Reiber and Felgenhauer ( R / F ) (Reiber, 1980; Reiber and Felgenhauer, 1987) formula is an empirically derived quantification of intrathecal IgG synthesis based on the differences in permeability between an intact and damaged BBB. All of the above calculations require determination of CSF as well as serum IgG and albumin concentrations. Schuller and Sagar (S/S) (Schuller and Sagar, 1983; Schuller et al., 1987) have attempted to simplify the calculations by eliminating serum albumin from their quantitative formula. For these authors an increase in CSF albumin (above normal values) signals a transudation of serum proteins, independent of serum protein concentration. They support their formula with a high correlation between the Tourtellotte formula and their calculations (Schuller and Sagar, 1983; Schuller et al., 1987). A fundamental assumption for using the Tourtellotte formula (Tourtellotte et al., 1980) in intrathecal IgG synthesis rate calculations is that the increase in CSF IgG concentration in the case of a damaged BBB is directly proportional to the increase in CSF albumin concentration. This assumption has been validated by a radioisotope two compartment study, in which radiolabeled IgG and albumin from pooled normal sera were followed from the blood to the CSF in clinical definite MS patients (Tourtellotte et al., 1980). The albumin leakage rate (AL), in mg/day, measures the quantity of albumin that crosses the BBB (Tourtellotte et al., 1989). Values of AL rate less than 75 m g / d a y (2 SD above mean of normal population) indicate a normal BBB and higher values indicate a damaged BBB. The effectiveness of the intrathecal IgG synthesis rate calculation depends on an accurate correction of an excess of IgG coming from blood and thus on a correct estimate of damage to the BBB. Another index of BBB damage is the CSF to serum albumin ratio (Ganrot and Laurell, 1974; Tibbling et al., 1977), called the albumin index (AI), which is used implicitly in the IgG index formula to correct for BBB damage. In this paper we investigated the changes of intrathecal IgG calculations mentioned above with re-

spect to increases in AL. One purpose was to determine which calculations make an appropriate correction for BBB damage as indicated by increasing AL values in chronic progressive MS patients. We established differences between formula calculations in AL subgroups of MS patients from three different clinical trials. We compared the frequency of occurrence of intrathecal IgG synthesis among the formulae, and determined correlations between the T formula and the other three formulae. Additionally, we compared the AL and AI formulae for BBB damage.

Materials and Methods Formulas Tourtellotte (T) intrathecal IgG synthesis rate in mg/day, formally called intra-blood-brain-barrier IgG synthesis rate (Tourtellotte et al., 1985a)

= [{ IgGcsv

IgG~ 369 ) - (AlbcsF

~[ IgGs ~ 1 Albs 230 )~ Albs ) X 0.43

×5 where CSF and serum (S) concentrations for both IgG and albumin (Alb) are in mg/dl; 369 is the serum/CSF ratio constant for IgG; 230 is the serum/CSF ratio constant for albumin; 0.43 is the molecular weight ratio of albumin to IgG; 5 is the daily CSF production in dl; the upper limit of normal is < 3.3 mg/day. Reiber and Felgenhauer ( R / F ) locally synthesized IgG fraction in m g / l (Reiber and Felgenhauer, 1987)

=

15 I

where concentrations are in mg/1 for CSF and g/1 for serum; 0.8, 15 and 1.8 are empirically derived constants for IgG from a general hyperbolic function valid for all immunoglobulin classes that describes the relationship between CSF/serum for immunoglobulin plotted on double log plots against CSF/serum for albumin; he upper limit of normal is ~ 0. Schuller and Sagar (S/S) intrathecal IgG synthesis in m g / l (Schuller et al., 1987)

where concentrations are in rag/1 for CSF and g/1 for serum; 30 rag/1 is the upper limit of normal for CSF IgG; 210 m g / l is the mean normal concentration of CSF albumin. Schuller and Sagar (Schuller et al., 1987)

187 indicate if CSF albumin is < 210, there is no transudation and only 30 should be deducted from CSF IgG. To avoid null values and maintain comparability with other equations, the restriction against negative values was eliminated and negative values for Schuller and Sagar were permitted. However, all values < 0 indicated no intrathecal synthesis. The constant 210 is different than the value of 240 indicated in Schuller et al. (1987) and was based on a personal communication from Schuller (1989); 60 is the equivalence of transudation for IgG; by definition the upper limit of normal is < 0. Tibbling and Link ( T / L ) IgG index (Tibbling et al., 1977) IgGcsF ] IgGs ] AlbcsF ] Albs ] where CSF units for IgG and albumin are identical ( m g / d l in our series), as are serum units for IgG and albumin (mg/dl), and the index values are dimensionless; upper limit of normal used in this study is < 0.7. Tourtellotte albumin leakage rate (AL) in m g / d a y (Tourtellotte et al., 1989)

AlbcsF

Albs ) 230 × 5

where CSF and serum values are in m g / d l and the constants 5 and 230 are identical to those for the Tourtellotte intrathecal IgG synthesis formula above; upper limit of normal is < 75 m g / d a y for an intact BBB. Albumin CSF to serum ratio, called the albumin index (AI) in this paper (Tibbling et al., 1977) Albcsv ] × 103 Albs ] a dimensionless value with upper limit of normal < 7.0.

expanded disability status scale (Kurtzke, 1983) (EDSS) in the last 6 - 8 months for the Azathioprine, or 12 months for the interferon-a and Cyclosporine studies, was an inclusion criterion for these trials. The patients had chronic progressive MS with generally high disability scores (the majority had EDSS > 5). Ten patients in the interferon-a study had less disabling forms.

CSF specimens All samples were drawn before drug treatment according to a standard protocol. IgG and albumin concentrations were quantified in fresh unfrozen a n d / o r frozen CSF and serum by electroimmunodiffusion (EID) (Tourtellotte et al., 1971). No differences were found between the frozen and unfrozen specimens (data not shown), which confirms a previous study (Tourtellotte et al., 1971). Statistical methods Data from the three studies were combined. We stratified the sample of MS patients based on increasing A L in m g / d a y into eight subgroups ( < - 2 5 m g / day, n = 2 9 ; > - 2 5 to < 0 m g / d a y , n = 1 0 2 ; > 0 to < 25 m g / d a y , n = 129; > 25 to < 50 m g / d a y , n = 115; > 50 to < 75 m g / d a y , n = 59; > 75 to < 150 mg/day, n = 7 2 ; > 1 5 0 to < 3 0 0 m g / d a y , n = 2 0 ; > 3 0 0 m g / day, n = 7). Pearson product moment correlations were used to analyze relationships among CSF measures and between A L and the various intrathecal IgG synthesis equations. Chi-square with Yate's correction for continuity were used to compare frequencies of abnormal findings among the formulas.

Results

Table 1 shows CSF and serum values for all patients. We found a statistically significant correlation TABLE 1 CSF and serum variables Variable CSF IgG (mg/dl)

Patients Pairs of CSF and serum specimens were obtained from 366 patients from a multicenter Cyclosporine T M A clinical trial (The Multiple Sclerosis Study Group, 1990), 68 patients from an Azathioprine clinical trial (Ellison et al., 1989), and 99 patients from an interferon-a clinical trial (Kastrukoff et al., 1990, 1991). All patients selected from the studies had clinically definite or laboratory-supported MS in the chronic progressive phase. An increase of one point on the Kurtzke

Serum IgG (mg/dl) CSF IgG/serum IgG ( × 10 -3) CSF albumin (mg/dl) Serum albumin (mg/dl) a Mean (standard deviation) b Range.

All cases n = 533 10.7 (6.5) a 2.2-42.6 b 1464 (366) 580--3140 7.4 (4.3) 1.9--29 32.0 (15.9) 12-210 5509 (547) 3780-7120

188 25.0

a.

~

" 140 *

120

"

t% 100 ~/~

200

0 0 :g

15.0

oll~

_.

-40 100

~

"

"

--m-

albu i

CSF

in

"20

5.0

0 -75

0

75

150

225

300

375

450

525

Albumin Leakage Rate (rag/day) 80

,o

+..°

1.6

,.4 f",,,

Q

s="

- 1.8

b.

# 40

;if

20

/

- ~$ 0

e. ~p

•

't'~' ,' ~

""

" 1.4

,

\-.

""''"I

~ TOURTELLO'Iq'E , o-i.-- REIBEPJFELGENHAMER ---O-- Schuller/Sagar - -e,- -Tibbling/Link

-20 -40 -75

}

" . S . ~ ....

"

0

i

i

75

150

225

'o.0

" 0.6 '.

300

0.4

t

,

375

450

0.2 525

Albumin Leakage Rate (rag/day)

Fig. 1. (a) Relationship of CSF IgG and albumin to trans-BBB albumin leakage rate. (b) Relationship between Tourtellotte, Reib e r / F e l g e n h a u e r , Tibbling/Link, and Schuller/Sagar intrathecal IgG synthesis equation values on the y-axes, and trans-BBB albumin leakage rate on the x-axis. Each data point represents the mean of chronic progressive MS patients with AL values within one of eight specified ranges: < - 2 5 ; - 2 5 t o < 0 ; 0 t o < 2 5 ; 2 5 to < 5 0 ; 5 0 t o < 75; 75 to < 150; 150 to < 300; > 300 m g / d a y .

between serum and CSF for both IgG (r = 0.30, P = 0.0001) and albumin ( r = 0.23, P = 0.0001). Fig. la shows the relationship between CSF IgG and albumin and trans-BBB AL. There was a progressive increase in CSF IgG with increasing A L values, with a steeper slope for normal A L values.

The relationship between intrathecal IgG synthesis by each of the formulas and trans-BBB A L is shown in Fig. lb and Table 2. In all cases the curves in Fig. lb were non-linear. For normal values of A L ( < 75 m g / day) there was a small positive correlation between A L and the T (r = 0.15, P < 0.001) and R / F (r = 0.10, P < 0.01) formulae (Table 2). There was a small negative correlation between A L and T / L (r = -0.10, P < 0.01), and no correlation between A L and the S / S formula. In all cases the significant correlations accounted for less than 2% of common variance. In contrast, for abnormal values of A L ( > 75 m g / d a y ) there was no correlation between A L and the T formula, but a modest negative correlation between A L and the R / F (r = -0.24, P < 0.05), T / L (r = -0.22, P < 0.01) and S / S (r = - 0 . 3 7 , P < 0.01) formulae. With all cases combined, the R / F formula showed no correlation with AL, while there was a negative correlation with A L for T / L and S / S and a positive correlation with A L for the T formula. Again, in all cases the common variance with A L was less than 4% for each formula. Fig. 2 a - c shows a comparison between intrathecal IgG synthesis rate from the T formula compared to each of the other three formulae. As indicated the T and R / F formula values are most highly correlated (r = 0.98, P < 0.0001), the T / L and T formulae correlate least (r = 0.82, P = 0.0001), with the S / S and T formulae showing an intermediate correlation (r = 0.92, P < 0.0001). Using their respective cutpoints to determine abnormal synthesis values, the majority of MS patients showed abnormal synthesis. Concordance of establishing a b n o r m a l / n o r m a l synthesis is also shown in Fig. 2, with concordance ranging from 95.1% for the R / F and T / L with the T formula, to 88.4% for the S / S and T formulae. Fig. 2d shows that the relationship between the A L and AI formulae for BBB leakage is nearly perfect

TABLE 2 Correlation between trans-BBB albumin leakage (AL) and CSF variables Variable

Normal AL n = 434 r (r 2)

Abnormal AL n = 99 r (r 2)

All cases n = 533 r (r 2)

CSF IgG Serum IgG C S F / S e r u m IgG CSF Albumin Serum Albumin C S F / S e r u m Albumin Tourtellotte Reiber/Felgenhauer Tibbling/Link Schuller/Sagar

0.24 0.04 0.28 0.92 0.08 0.99 0.15 0.10 -0.10 0.05

0.40 0.01 0.48 0.99 0.00 0.99 0.07 -0.24 -0.22 -0.37

0.43 0.02 0.47 0.99 0.08 1.00 0.19 - 0.01 -0.16 -0.16

(0.06) (0.00) (0.08) (0.85) (0.01) (0.99) (0.02) (0.01) (0.01) (0.00)

r = Pearson product moment correlation coefficient. * P < 0.001; ** P < 0.01; * * * P < 0.05.

a * * * ** ** **

(0.16) (0.00) (0.23) (0.99) (0.00) (0.99) (0.00) (0.06) (0.05) (0.13)

" * * * *** *** *

(0.18) (0.00) (0.22) (0.98) (0.01) (0.99) (0.04) (0.00) (0.03) (0.02)

* * * * * * *

189 2b. Relationship between Tibbling-Link IgG Index and Tourtellotte Equation

2a. Relationship between Reiber-Felgenhauer and Tourtellotte Equations 125.0

in=s33, r=0.N, r.-0.95, p=o.0o011

~" 10010 o

.r.

-- d 7S.0

o1 '3

5.0 - -

•

,P;

,

Concordance = 95.1% ] corrected Chl-square=-259 p=O.O001

4.0 - -

•

• •

S "

2.0 4abnormal n=476 - 89.3%

n:17 0.0

both ~ e ~ e . . . . . I:'-I "

":~.'..,I....

-25.0

-25.0

0.0

•, n:s

•

I Concordance = 95.1% I I v a t . ' COl'r.ted Chl-square--.239[

,...

I., ....

, ,.:.o..oo.o? ....

,.

25.0 50.0 75.0 100.0 Intrethecel IgG S y n t h e s i s Rate (mg/day)

both 1.0 normal

I..,

125.0

n=38

0. 0

~

e ~ l . . . .

~".

l.l&~',,~_ •

ZI n-12 , -25.0

150.0

.',r...+."

I •

~

•

"

.

" "" •

"

•

bo,,

abnormal n1471 ; 88.4%

•

n=14 it I . . . .

0.0

i

. . . .

I

. . . .

i

. . . .

i

. . . .

25.0 S0.0 75.0 100.0 Intrathecal IgG S y n t h e s i s Rate (mg/day)

I

. . . .

125.0

40.0

400.0 I

In=533,

r=0.92,

r~0.85,

n=533,

p=o.oo01 I

r_-0.99,

r2=0.99,

300.0

I" J

32.0 •

•

ell•

•

J

28.0

.:....,.:'"

~2oo.o

p

p=0.0001]

36.0

.. '

i

150.0

2d. Relationship between Albumin Index and T-BBB Albumin Leakage Rate

2c. Relationship between Schuller-Sagar and Tourtellotte Equations

f

"~C 2 4 . 0

.~ E

o

p=o.0oolL

rl=0.s7,

'o c

25.0

n-

z

Yalas'

r=0.82,

50 0

'3

m •~

--

In.-X3$,

.'•

(3

~

0.0

.. • •

lOO.O - -

,t

2O.0

/

f

/

•

j~ 16.0

j.../

+ n--21

8

"

.bnorm.,

12.0

n=0

,J o.o

-

1,1.t bo, ,:.n.,.. . .h"."T i';"

8.0

-21.0

0.0

.

.

.

.

.

.

,

.

20.0 IO.O 70.0 100.0 Intrathecal IgG Synthesis Rate (mg/dayl

.

.

.

120.0

Concordance = 97.7% n=4 Yales' corrected Chi-squmre=452 .on normal-= p=0.0001 n=420 0.0 .. • i,. , i . . , i,, ,I, , , i . . . i . . . i ,,. l., 11 ,., 100.0 300.0 000.0 700.0 900.0 -100.0 Trans-Blood-Braln-Barrler Albumin Leakage Rate (rag/day)

4.0 -."j

I Conoordance = 88.4% ,--.rr...,,;o ooo,C+"r'="A .

~100~ 0

e J •Both ,a,n~mal n=0S

~t~

!

100.0

I

Fig. 2. (a-e) Relationship between TourteUotte equation values (on x-axis) and Reiber/Felgenhauer (a) and Schuller/Sagar (c) intrathecal IgG synthesis equation values (on left y-axis), and Tibbling/Link (b) IgG Index values (on right y-axis). The Reiber/Felgenhauer and Schuller/Sagar values were scaled by 0.5 before plotting to emphasize their similarity with Tourtellotte values for normal albumin leakage rates ( > 75 mg/day). d shows relationship between albumin leakage rate and albumin index (see text for equations) as measures of trans-BBB albumin leakage.

(r = 0.99, P < 0.0001), with a concordance of 97.7% in determining n o r m a l / a b n o r m a l leakage.

Discussion The Tourtellotte and R e i b e r / F e l g e n h a u e r formulae are based on the assumption that the extent of the leakage of proteins from blood to CSF is a function of both the source concentrations and the BBB permeability. Therefore, IgG measured in CSF is the result of intrathecal IgG synthesis plus the IgG transudate from serum. When the BBB is damaged, both IgG and albumin cross to the CSF compartment. Since albumin

is not produced in the CNS (Schliep and Felgenhauer, 1978), excess albumin in CSF must come from serum, and it is this fact that makes it a good marker of protein transudation. In order to quantify the leakage of IgG across the BBB, we established its relationship with albumin leakage. The correction factors used in our formula that are subtracted from the IgG (serum IGG/369) and albumin (serum albumin/230) are based on the normal leakage of IgG and albumin into the CSF from the blood and have a good concordance with the results of an in vivo isotope test (Tourtellotte et al., 1980). Link and colleagues (Lefvert and Link, 1985) did not find a linear correlation between serum and CSF proteins in

190 a referential population. This lack of correlation does not mean that the proteins of the CSF and serum do not interact with each other (Felgenhauer, 1982, 1984; Tourtellotte et al., 1985b). In this group of MS patients we report significant, but low correlations between CSF and serum values for both IgG and albumin. The results of intrathecal IgG synthesis rate calculations by the various formulae had an excellent correlation with each other and good concordance in determining abnormal synthesis. However, there were definite differences in the relationship between IgG synthesis and AL values, particularly for abnormal AL indicating a break in the BBB. From our data, the R / L , S / S and T / L formulae all showed significant negative correlations with increasing AL, particularly the S / S formula. It is unreasonable to believe that in chronic progressive MS patients intrathecal IgG synthesis should be significantly lower for increasingly abnormal values of AL, which indicates increasingly more severe breakage of the BBB. How the S / S formulae overcompensates for high values of AL is unclear, since intrathecal IgG synthesis rate calculated by the Schuller formula is simply the CSF IgG concentration minus a corrective constant. Schuller's corrective factor is based on normal albumin levels in the CSF. He believes this discount, which is independent of AL, is sufficient because he has found no linear correlation between CSF and serum albumin levels (Schuller et al., 1987). He misinterprets this to mean that changes in serum protein concentrations do not affect transudation into the CSF. However, a relationship between serum and CSF proteins has been shown (Felgenhauer, 1982, 1984; Lolli et al., 1989; Tourtellotte et al., 1985b), which we also report in this study. The T / L index corrects the CSF IgG increase due to increases in serum protein levels. The T / L index had a tendency to overcompensate when AL values were more than 75 m g / d a y (BBB damage). When AL values were very high (over 200), the T / L index approached normal values. The fact that these high AL data were measured in MS patients leads us to believe that the correction was excessive. There is no reason to expect that IgG synthesis should be lower in MS patients with high AL. The Reiber equation and our formula corrected the AL, particularly when BBB damage occurred. When compared with each other, the formulae exhibited a good correlation overall (r = 0.98), which was obtained even when the BBB was damaged (r = 0.92). The only discrepancy was that the value calculated by the Reiber formula had a tendency to decrease for high AL levels. Since BBB damage might be expected in more advanced, or more active MS disease, increased intrathecal IgG synthesis would be an expected finding. It should be noted that the results and conclusions in this paper cannot be generalized outside the popula-

tion of study. Based on their comparisons among formulae in other neurological disorders, Ohman et al. suggested that the R / F and a modified form of the T / L may provide an overall more sensitive and specific measure of intrathecal IgG synthesis than other formulae (0hman et al., 1992). However, Ohman et al. did not specifically address the influence of high AL (or AI) on sensitivity and specificity. It may be that in the presence of severe BBB damage, an intrathecal immune response is an expected consequence regardless of the neurological disease, so that formulae that do not adequately correct for albumin leakage at high values may yield false negative results. Likewise, assays for oligoclonal bands may be compromised in cases of severe BBB damage by the presence of an overwhelming influx of IgG from the serum. We feel that the issue of true intrathecal IgG synthesis in the event of high AL in other neurological diseases remains open. The excellent concordance among the formulae could be related to the high number of cases, the certainty of MS diagnosis, the homogeneity of patients, and the validity of the underlying concept of intrathecal IgG synthesis as a sine qua non of chronic progressive MS. The fact that there existed no significant correlation between an IgG/albumin rate quotient and AL supports the idea that the relationship between IgG and CSF/serum albumin ratios is independent of BBB damage. Thus, at least in chronic progressive MS, there is no need to change permeability constants in the presence of BBB damage as recommended by Peter and Bowman (1992). In conclusion, we have established a significant relationship between the AL and IgG concentration and the necessity for a valid correction of IgG transudate when AL is higher than 75. In chronic progressive MS the best correction can be expected from the Tourtellotte formulae. The other three formulae can provide sensitive measures of intrathecal IgG synthesis rate; however, all have a tendency for false negatives when the BBB is compromised.

Acknowledgements Specimens were obtained from the National Neurological Research Specimen Bank (an anatomical donor program and collection of cryopreserved specimens for neurosciences), located at VAMC West Los Angeles, Los Angeles, CA 90073, which is sponsored by NINC D S / N I M H , National Multiple Sclerosis Society, Hereditary Disease Foundation and the Dystonia Medical Research Foundation. This work was supported in part by grants from the Veterans Health Services and Research Administration, Department of Veterans Affairs (Merit Review funding) and from the Sandoz Company Study # 334/335.

191

Multiple Sclerosis Study Group: Sandoz Research Institute, Sandoz Pharmaceutical Corp, East Hanover, N J: George Belenduk, MD, PhD; Barbara Kasper, RN; Diane Klatzman, RN; William Mietlowski, phD; Suzanne Solch, RN. Rocky Mountain Multiple Sclerosis Center, University of Colorado Health Sciences Center, Denver, CO: Gary Franklin, MD, MPH; Jack Burks, MD; Lorene Nelson, MS; Carolyn Wangaard, RN, CANP. Neuroimmunology Branch, National Institutes of Health, Bethesda, MD: Henry McFarland, MD; Andrew Goodman, MD; Dale McFarlin, MD; Helen Krebs, RN; Heidi Maloni, RN; Joe Debronozo, PhD. Albert Einstein College of Medicine, Bronx, NY: Labe Scheinberg, MD; Ute Traugott, MD; Mindy Aisen, MD; Kate Robbins, CCOT. Arizona Health Sciences Center, Tucson, AZ: William Sibley, MD; Jose Laguna, MD; Joan Laguna. Washington University School of Medicine, St Louis, MO: John Trotter, MD; David Cliford, MD; Larry Smith, MD; Jane Mcinnis, RN. University of Chicago, Chicago, IL: Barry Arnason, MD; Raymond Roos, MD; Anthony Reder, MD; Jack Antel, MD; Mark Aguis, MD; Roberta Martia, RN. Duke University Medical College, Durham, NC: Barrie Hurwitz, MD; Steven Greenburg, MD; Louis Fredane, MD; Rebecca Herbstreith, RN; Jean Hurwitz. University of Maryland School of Medicine, Baltimore, MD: Kenneth Johnson, MD; Hillel Panitch, MD; Carol Lee Koski, MD; Paul Fishman, MD; Sue Haley, RN. University of Utah Medical Center, Salt Lake City, UT: Jack Petajan, MD, PhD; Patrick Bray, MD; John Rose, MD; David Thurman, MD; William Galster, MS. Neurology and Research Services, VAMC W. Los Angeles, Wadsworth Division, Los Angeles, CA: Wallace Tourtellotte, MD, PhD; R.W. Baumhefner, MD; George Ellison, MD; Lawrence Myers, MD; Karl Syndulko, PhD; Lavona Newton, RN. University of Texas Health Sciences Center at Houston, TX: Jerry Wolinsky, MD; E. Simon Sears Jr., MD; Avindra Nath, MD; Catherine Weisbrodt, RN. Alpha Interferon Study Group: Department of Medicine, Health Sciences Centre Hospital, University of British Columbia, Vancouver, BC: L.F. Kastrukoff, MD, F.R.C.P.; J.J. Oger, MD; D. Paty, MD. Azathioprine Study Group: Neurology and Research Services, VAMC W. Los Angeles, Wadsworth Division, Los Angeles, CA: R.W. Baumhefner, MD. Department of Neurology, University of California at Los Angeles School of Medicine, LOs Angeles, CA: G.W. Ellison, MD; L.W. Myers, MD.

References Brightman, M.W. (1965) The distribution within the brain of ferritin injected into cerebrospinal fluid compartments. II. Parenchymal distribution. Am. J. Anat. 117, 193-220. Ellison, G.W., Myers, L.W., Mickey, M.R., Graves, M.C., Tourtellotte, W.W., Syndulko, K., Holevoet-Howson, M.I., Lerner, C.D., Franc, M.V. and Pettler-Jennings, P. (1989) A placebo-controlled, randomized, double-masked, variable dosage, clinical trial of azathioprine with and without methylprednisolone in multiple sclerosis. Neurology 39, 1018-1026. Felgenhauer, K. (1982) Differentiation of the humoral immune response in inflammatory diseases of the central nervous system. J. Neurol. 228, 223-237. Felgenhauer, K. (1984) Protein size and cerebrospinal fluid composition. Klin. Wschr. 42, 1158-1164. Ganrot, K. and Laurell, C.B. (1974) Measurement of IgG and albumin content of cerebrospinal fluid, and its interpretation. Clin. Chem. 20, 571-573. Kastrukoff, L.F., Oger, J.J., Hashimoto, S.A., Sacks, S.L., Li, D.K., Palmer, M.R., Koopmans, R.A., Petkau, A.J., Berkowitz, J. and Paty, D.W. (1990) Systemic lymphoblastoid interferon therapy in chronic progressive multiple sclerosis. I. Clinical and MRI evaluation. Neurology 40, 479-486. Kastrukoff, L.F., Oger, J.J., Tourtellotte, W.W., Sacks, S.L., Berkowitz, J. and Paty, D.W. (1991) Systemic lymphoblastoid interferon therapy in chronic progressive multiple sclerosis. II. Immunologic evaluation. Neurology 41, 1936-1941. Kurtzke, J.F. (1983) Rating neurologic impairment in multiple sclerosis: An expanded disability status scale (EDSS). Neurology 33, 1444-1452. Lefvert, A.K. and Link, H. (1985) IgG production within the central nervous system: a critical review of proposed formulae. Ann. Neurol. 17, 13-20. Lolli, F., Halawa, I. and Link, H. (1989) Intrathecal synthesis of IgG, IgA, IgM and IgD in untreated multiple sclerosis and controls. Acta Neurol. Scand. 80, 238-247. Myers, C.D. (1991) Role of B cell antigen processing and presentation in the humoral immune response. FASEB J. 5, 2547-2553. Ohman, S., Ernerudh, J., Forsberg, P., Henriksson, A., von Schenck, H. and Vrethem, M. (1992) Comparison of seven formulae and isoelectric focusing for determination of intrathecally produced IgG in neurological disease. Ann. Clin. Biochem. 29, 405-410. Peter, J.B. and Bowman, R.L. (1992) Intra-blood-brain-barrier synthesis of lgG: Comparison of IgG synthesis formulas in a computer model and in 1,629 consecutive specimens. Neurology 42, 510-515. Prineas, J.W. (1985) The neuropathology of multiple sclerosis. In: J. Koetsier (Ed.), Handbook of Clinical Neurology, Elsevier, Amsterdam, 3/47, pp. 213-257. Reiber, H. (1980) Evaluation of lgG synthesis in the central nervous system. In: H. Bauer, S. Poser and G. Ritter (Eds.), Progress in Multiple Sclerosis Research, Springer, Berlin, pp. 154-156. Reiber, H. and Felgenhauer, K. (1987) Protein transfer at the blood cerebrospinal fluid barrier and the quantitation of the humoral immune response within the central nervous system. Clin. Chim. Acta 163, 319-328. Schliep, G. and Felgenhauer, K. (1978) Serum-CSF protein gradients, the blood-CSF barrier and the local immune response. J. Neurol. 218, 77-96. Schuller, E. and Sagar, H.J. (1983) Central nervous system IgG synthesis in multiple sclerosis. Acta Neurol. Scand. 67, 365-371. Schuiler, E.A.C., Benabdailah, S., Sagar, H.J., Reboul, J.A.M. and Tompe, L.C. (1987) IgG synthesis within the central nervous system. Arch. Neurol. 44, 600-604. The Multiple Sclerosis Study Group (1990) Efficacy and toxicity of cyclosporine in chronic progressive multiple sclerosis: A random-

192 ized, double-blind, placebo-controlled clinical trial. Ann. Neurol. 27, 591-605. Tibbling, G., Link, H. and Ohman, S. (1977) Principles of albumin and IgG analyses in neurological disorders. Scand. J. Clin. Lab. Invest. 37, 385-390. Tourtellotte, W., Tavoata, B., Parker, J.A. and Comiso, P. (1971) Cerebrospinal fluid electroimmunodiffusion. An easy, rapid, sensitive, reliable, and valid method of the simultaneous determination of immunoglobulin-G and A. Arch. Neurol. 25, 345-350. Tourtellotte, W.W., Potvin, A.R., Fleming, J.O., Murthy, K.N., Levy, J., Syndulko, K. and Potvin, J.H. (1980) Multiple sclerosis: Measurement and validation of central nervous system IgG synthesis rate. Neurology 30, 240-244. Tourtellotte, W.W. and Walsh, M.J. (1984) Cerebrospinal fluid profile in multiple sclerosis. In: C.M. Poser, D.W. Paty, L. Scheinberg, W.I. McDonald and G.C. Ebers (Eds.), The Diagnosis of Multiple Sclerosis, Thieme-Stratton, New York, pp. 165-178.

Tourtellotte, W.W., Staugaitis, S., Walsh, M.J., Shapshak, P., Baumhefner, R.W., Potvin, A.R. and Syndulko, K. (1985a) The basis of intra-blood-brain barrier IgG synthesis. Ann. Neurol. 17, 21-27. Tourtellotte, W.W., Staugaitis, S.M., Walsh, M.J., Shapshak, P., Baumhefner, R.W. and Potvin, A.R. (1985b) The science of intra-blood-brain barrier IgG synthesis. In: H. Peeters (Ed.), Protides of the Biological Fluids. Proceedings of the 32 Colloquium, Pergamon Press, New York, pp. 161-164. Tourtellotte, W.W., Shapshak, P., Osborne, M.A., Rubinshtein, G., Lee, M. and Staugaitis, S.M. (1989) New formula to calculate the rate of albumin blood-brain-barrier leakage. Ann. Neurol. 26, 176. Whitaker, J.N. (1985) Quantitation of the synthesis of immunoglobulin G within the central nervous system. Ann. Neurol. 17, 11-12.