632

INCREASED LEVELS OF PLASMA ANAPHYLATOXINS IN SYSTEMIC LUPUS ERYTHEMATOSUS PREDICT

FLARES OF THE DISEASE AND MAY ELICIT VASCULAR INJURY IN LUPUS CEREBRITIS

PATRICIA HOPKINS, H. MICHAEL BELMONT, JILL BUYON, MARK PHILIPS, GERALD WEISSMANN, and STEVEN B. ABRAMSON

We measured levels of complement anaphyla- toxin split products, C3a and C5a, in the circulation of patients with systemic lupus erythematosus (SLE). In 23 SLE patients who were followed serially, the mean C3a value was 179 ng/ml during stable disease and 550 ng/ml during a disease flare. In 10 patients, C3a levels pre- dicted disease activity, with the C3a value rising from a mean of 183 ng/ml at a time of stable disease to a mean of 242 ng/ml 1-2 months prior to a clinical exacerbation of disease. The mean C3a level in 5 patients with acute dysfunction of the central nervous system (CNS) was 1,297 ng/ml, which is significantly higher than that observed in patients with active disease but without CNS involvement (P < 0.01). C5a levels were also signifi- cantly elevated in 4 patients with acute CNS disease. Pathologic specimens from 2 patients who died during an acute lupus flare revealed neutrophils occluding the cerebral and intestinal vessels. Fluorescein angiography in a patient with CNS lupus revealed vasoocclusive retinopathy. In 5 of 7 SLE patients who were pregnant, C3a levels were elevated, with a group mean value of 310 ng/ml. There was a negative correlation (r = - 0.59) between C3a and C3 levels in pregnant patients with SLE, and this finding is consistent with comple-

~~

From the Division of Rheumatology, Department of Medi- cine, New York University Medical Center, New York, New York.

Dr. Abramson’s work was supported by NIH grant AM- 01431-30. Dr. Weissmann’s work was supported by NIH grants AM-11949, HL-19721, and AM-01431-30.

Patricia Hopkins, MD; H. Michael Belmont, MD; Jill Buyon, MD; Mark Philips, MD; Gerald Weissmann, MD; Steven B. Abramson, MD.

Address reprint requests to Steven B. Abramson, MD, Department of Medicine, Division of Rheumatology, NY U Medical Center, 550 First Avenue, New York, NY 10016.

Submitted for publication June 15, 1987; accepted in re- vised form November 4, 1987.

ment activation as the cause of decreasing C3 levels. We suggest that serial measurements of C3a can predict flares of disease in lupus patients and can demonstrate complement activation during pregnancy in women with SLE. In addition, release of C3a and C5a (mediators of inflammation) into the circulation may elicit vascular injury, particularly in patients with lupus cerebritis.

The activation of complement by immune com- plexes deposited in tissues plays a central role in the manifestations of systemic lupus erythematosus (SLE) (1). The consumption of complement components is reflected by a decrease in serum levels of C3 and C4 in most patients with active disease (2,3). However, since the synthesis of both C3 and C4 increases during periods of disease activity, the serum levels of these proteins may be normal despite the accelerated con- sumption (43). Conversely, decreased synthesis, he- reditary deficiencies of individual components, as well as the increased extravascular distribution of comple- ment proteins have been reported in patients with SLE (6,7). The decreased serum complement levels in these patients may lead to the mistaken conclusion that excessive complement activation is ongoing.

To define more precisely the role of comple- ment activation with respect to the activity of SLE, we (8) and others (9,lO) have measured circulating levels of complement degradation products. Levels of plasma C3a and the serum complement attack com- plex, SCSb-9, were shown to be more sensitive indi- cators of disease activity in SLE than either the C3 level or the C4 level (8-10). We report here a prospec- tive study in which we examined the clinical utility of serial measures of complement activation products in plasma. Our results indicate that elevations of plasma

Arthritis and Rheumatism, Vol. 31, No. 5 (May 1988)

C3a AND C5a IN SLE 633

1,000 c - E 2 800

600

c u

v) Q) -0

rF: 400 0

2173 1,920

0

b

a

:

Normals Stable Flare

Figure 1. Plasma levels of C3a in normal control subjects and in 23 patients with systemic lupus erythematosus who were followed serially during 2 phases of disease activity (stable and disease flare). The mean C3a level for this group of patients was significantly higher at the time of the disease flare than at the time of stable disease (P < 0.oOOl). Bars show group means; broken line shows normal mean + 2 SD.

C3a levels may precede other serologic or clinical evidence of an impending disease flare. Moreover, the elevation of this split product signals complement acti- vation in women with SLE who are pregnant. We also

present data and histopathologic evidence suggesting that elevated levels of C3a and C5a in plasma contribute to the vascular injury observed in patients with acute involvement of the central nervous system (CNS).

PATIENTS AND METHODS Patient and control populations. Plasma was obtained

from 40 patients who fulfilled the American Rheumatism Association revised criteria for SLE (1 1). These patients received medical care in the outpatient clinics and the inpatient services of the New York University Medical Center or the Hospital for Joint Diseases Orthopaedic Insti- tute. Blood samples were also drawn from 17 normal volun- teers and from 17 NYU Medical Center patients with renal disease that was not due to SLE.

Patients with SLE were evaluated by their physi- cians and classified as having stable or active disease. Patients were considered to have stable disease if they were taking no medication and were clinically asymptomatic or if their symptoms were controlled with a medical regimen of less than 0.5 mg/kg of oral prednisone each day. Patients were considered to have active disease if the following signs or symptoms were present: fever, serositis, arthritis, typical dermatitis, thrombocytopenia, nephritis, or CNS dysfunc- tion. Acute CNS dysfunction was defined as altered mental status, stupor, or coma resulting from no identifiable cause other than SLE.

Twenty-three patients were evaluated during at least 2 phases of disease activity (active and stable). We also evaluated 10 patients whose C3a values had been measured during a time of stable clinical disease, 3-6 months and 1-2 months prior to a disease flare, as well as during a disease flare.

Pregnant patient population. Plasma was obtained from 16 healthy pregnant women who were receiving fol-

Table 1. stable disease and at disease flare*

Prospective determinations of C3a desArg and C3 levels in 10 patients with systemic lupus erythematosus, during months of

~ ~ ~ ~ ~ ~~~~~

Stable disease, 3-6 Stable disease, 1-2 mos. before flare mos. before flare Disease flare

Patient C3a C3a C3a Feature at number desArg c 3 desArg c 3 desArg c 3 disease flare

1 197 76 253 66 318 69 Nephritis 270 I05 414 46 Nephritis 2 197 1 I8

3 215 90 275 97 763 36 Nephritis 4 182 120 233 80 325 70 Nephritis 5 124 46 I84 32 704 46 ARDS/vascular occlusion 6 240 116 324 100 695 46 Cutaneous vasculitis 7 137 97 197 108 408 83 Pericarditis 8 153 I68 200 157 347 117 Nephritis 9 152 110 229 77 560 94 Nephritis 10 227 1 24 253 140 486 125 Pencarditis

Mean ? SD 183 f 40 107 f 32 242 2 42t 96 t 36 502 2 169$ 73 +- 31

* Normal C3a desArg value 5200 ng/ml; normal C3 value 270 mg/dl. ARDS = adult respiratory distress syndrome. t P < 0.01 versus value found at 3-6 months before disease flare. $ P < 0.01 versus values found at 3-6 months and at 1-2 months before disease flare.

634 HOPKINS ET AL

A I 300

v c" 250 \

0

L- O Y, Q) U

8 0

0 0

8 i 0

0 0

0-1.5 1.6-4 >4 Creat in ine ( mg %)

Figure 2. Plasma levels of C3a in 17 patients with renal disease not related to systemic lupus erythematosus. C3a levels in these indi- viduals were, with 1 exception, 200 ng/ml or below. Broken line shows normal mean + 2 SD.

lowup care in the outpatient obstetric clinic. A detailed medical history, including history of hypertension, protein- uria, gestational diabetes, or preeclampsia, was obtained. Patients were excluded from the study if they were taking medications for concurrent medical conditions. Seven nor- mal pregnant women with preeclampsia were also evaluated. Preeclampsia was considered to be present if at least 2 of the following developed during pregnancy: diastolic blood pres- sure more than 80 mm Hg, increase in serum uric acid level, proteinuria, peripheral edema, or decreased creatinine clear- ance. Seven SLE patients were followed serially throughout pregnancy.

Determination of complement levels by radioimmuno- assay or immunodiffusion. Samples were evaluated for C3a and C5a by radioimmunoassay (kits generously provided by The Upjohn Co., Kalamazoo, MI). All plasma samples were prepared with 10 mM sodium EDTA and centrifuged at 1,600 revolutions per minute for 10 minutes at 4°C. All samples were stored at -70°C within 30 minutes of venipuncture. Assays were performed every 4-6 weeks. The radioimmu- noassays are specific for both of the native anaphylatoxins, as well as their stable derivatives C3a desArg and C5a desArg (12). The results of assays for C5a and 2 of the 5 results of assays for C3a in patients with CNS disease have been previously published (8). All assays were performed in duplicate. The upper limit of normal for the C3a value is 200 ng/ml, which is 2 SD above the mean level found in our normal control population (n = 35). Interassay and intra- assay variation of paired specimens was less than 10% and less than 15%, respectively. Total C3 and C4 levels were determined by immunodiffusion (13). The mean values for C3a were compared by the Mann-Whitney U test. C3a and

C3 values in pregnant patients with lupus were correlated using the Spearman rank correlation method.

RESULTS Plasma C3a levels during active and stable SLE.

Plasma was obtained from 17 normal controls and from 23 SLE patients who had been followed prospectively. Each patient was studied during both a stable and an active period of disease. As shown in Figure 1, levels of C3a were at or under 200 ng/ml in plasma from normal subjects (mean t SD 118 2 20 nglml). In SLE patients with stable disease, the mean level of C3a was elevated above the mean found in controls (179 ? 57 ng/ml) (P < 0.01). During a period of disease flare, plasma levels of C3a rose in each of the patients (group mean 550 ? 3 11 ng/ml). The difference in these levels during active versus stable disease periods was highly statistically significant (P < 0.OOOl). Thus, in individual SLE pa- tients who were assessed serially, plasma levels of C3a increased markedly during a disease flare.

Findings from the prospective study. To deter- mine whether a rise in plasma C3a precedes (and therefore predicts) clinical exacerbations of disease, 10 patients were followed prospectively for at least 6 months until a disease flare occurred. Plasma levels of C3a had been assessed in these patients when the disease was considered clinically and serologically sta- ble. As shown in Table 1 , the mean +- SD plasma level of C3a in this group of patients 3-6 months before a disease flare was 183 ? 40 ng/ml. Serum C3 levels (by immunodiffusion) were normal in all but patient 5. One to two months prior to disease flare, the C3a levels rose in each patient studied; typically, these levels rose to 200 ndml or higher (mean 2 SD 242 ? 42). Compared with the mean value 3-6 months prior to disease flare, this change was statistically significant (P < 0.01).

Two months prior to disease flare, the serum C3 levels had fallen in 7 of the 10 patients; however, the value was below the normal range in only 2 individu- als, and the decline for the population was not signif- icant. All 10 patients were considered by their physi- cians to have clinically stable disease at this time, and in no patient was an increase in therapy instituted based on the data available. A disease flare ensued in each of these patients, and an increase in prednisone to dosages above 40 mg/day or the addition of an immunosuppressive agent, such as azathioprine, was required. At the time of the disease flare, the mean +. SD plasma C3a level for this group of patients had risen to 502 ? 169 ng/ml (P < 0.01 versus pre-flare

C3a AND C5a IN SLE 635

levels). Serum C3 values, as detected by immunodif- fusion, remained within the normal range at the time of disease exacerbation in 5 of the 10 patients.

Patient 5 had hypocomplementemia throughout the period of observation, even when the disease was considered clinically stable. In contrast, the plasma C3a level rose progressively from a normal value of 124 ng/ml when the patient was clinically stable to 704 nglml at the time of disease flare. Thus, in this patient with chronic hypocomplementemia, in whom de- creased synthesis of complement proteins might have been a confounding variable, a rise in the level of the split product C3a proved to be a useful serologic indicator of complement activation and a predictor of disease exacerbation.

Renal function and C3a. Since the C3a content increased with exacerbations of disease, we deter- mined whether a deterioration in renal function alone would account for this observation. Plasma C3a levels were therefore measured in patients who had renal disease that was not related to SLE. The patients evaluated had stable renal function and the following diagnoses: amyloidosis (n = 2), minimal change dis- ease (n = 4), membranous glomerulonephritis (n = 3), diabetic nephropathy (n = 5) , and endstage renal disease (receiving dialysis) (n = 3). Plasma from the patients receiving dialysis was obtained immediately before the dialysis. As shown in Figure 2, the C3a levels in patients with renal insufficiency unrelated to SLE or other immune complex deposition diseases were, with one exception, below 200 nglml. C3a levels were also in the normal range in 6 control patients with significant proteinuria (range 2-1 1 gm of protein in 24 hours) in the absence of a rise in serum creatinine (data not shown).

These data indicate that renal disease alone, particularly a decline in creatinine clearance, is not sufficient to explain elevations of plasma C3a levels in patients with active SLE. Such elevations must there- fore require accelerated degradation of intact C3 in the plasma or tissues of these patients.

Circulating anaphylatoxins in acute CNS lupus. Increased plasma content of C3a was found in all SLE patients with active disease. However, when patients were classified with respect to organ system involve- ment, we noted that patients with acute CNS lupus represented a distinct subgroup. As shown in Figure 3, the mean 2 SD C3a level in 5 patients with CNS lupus was 1,297 f 667 ng/ml, which was significantly higher than that in patients with active SLE who did not have CNS involvement. Acute cerebral dysfunction was

also accompanied by an increase in plasma C5a levels, which was found in all such patients (mean 72.0 ng/ml). In contrast, C5a levels were normal in the majority of patients with active disease that did not involve the central nervous system (mean 10.2 ng/ml). This level was significantly lower (P < 0.01) compared with that found in patients with acute CNS disease (Figure 3).

Specimens for histologic examination were ob- tained from 2 of the patients and angiography of the eyes was performed in a third patient. The findings in these 3 patients support the possibility that CNS lupus may result from the intravascular activation of inflam- matory cells. Figure 4 shows a specimen of brain obtained postmortem from a 36-year-old woman with a history of fever, seizures, and coma. Her course was complicated by the development of the adult respira- tory distress syndrome (ARDS). At autopsy, the pulmonary findings were those of classic ARDS: intra- alveolar hemorrhage, neutrophilic infiltration, and lining of alveolar walls with thick hyaline membranes. The predominant neuropathologic lesion observed was microinfarction due to thrombi occluding the small

Non-CNS CNS Non-CNS CNS Disease Disease Disease Disease

Figure 3. Plasma C3a and C5a levels in systemic lupus erythema- tosus patients with or without central nervous system (CNS) in- volvement. The plasma C3a value was elevated in all patients with active disease, and the mean C3a value was significantly elevated in patients with acute cerebral dysfunction (P < 0.01). The plasma CSa level was normal in most of the patients with non-CNS disease, but it was significantly elevated in patients with active lupus who had CNS involvement (P < 0.01). Bars show group means; broken line shows normal mean + 2 SD.

636 HOPKINS ET AL

A B Figure 4. Specimen of brain obtained postmortem from a 36-year-old woman with adult respiratory distress syndrome and acute central nervous system lupus. A, A cerebral infarction subtended by a vessel that is occluded by an aggregate of polymorpho- nuclear leukocytes (arrow). B, Higher magnification of the occluded vessel.

vessels. Careful inspection of the specimens revealed cerebral infarcts subtended by vessels that were oc- cluded not only by fibrin thrombi, but also by aggre- gates of polymorphonuclear leukocytes (Figure 4).

The other specimen was obtained from a 30- year-old SLE patient with a history of fever, disorien- tation, and abdominal pain. Her condition had been diagnosed as CNS lupus and mesenteric vasculitis. Despite high-dose corticosteroid therapy, the patient developed stupor and then coma. On the fifteenth day of hospitalization, she died as a result of widespread intestinal infarction. The brain was not available for postmortem study; however, the infarcted jejunum was examined histologically, and it revealed submu- cosal edema and multiple small vessels occluded by leukothrombi (Figure 5) . Leukothrombosis (in the absence of vasculitis) was the sole pathologic abnor- mality noted which could account for the extensive infarction of the bowel. The third patient had a history of CNS lupus, and had presented with a complaint of

blurred vision. Results of a routine ophthalmologic examination were normal, and fluorescein angiogra- phy was performed. The angiogram revealed areas of pericapillary leakage of dye, as well as vasoocclusive retinopathy (Figure 6).

Pregnancy and plasma C3a content. In lupus patients who are pregnant, it has not been clear whether a decline in total C3 and C4 contents follows activation of complement by the disease or whether it is due to decreased synthesis of complement protein. Figure 7 shows the circulating levels of total C3 and C3a in SLE patients and control subjects who were pregnant. As expected, in healthy pregnant women, the total C3 value was in the high normal range, and except for 1 subject, these women had normal levels of plasma C3a. In subjects with preeclampsia (non-SLE), the total C3 content was also found to be in the high normal range; again, with 1 exception, the C3a value was below 200 ng/ml in these women.

Seven pregnant SLE patients were studied.

C3a AND C5a IN SLE 637

A B Figure 5. Specimen of jejunum obtained postmortem from a 30-year-old woman with small bowel infarction and central nervous system lupus. A, Submucosal edema and multiple small vessels occluded by leukothrombi (arrow) are seen. B, Higher magnification of polyrnorphonuclear leukocyte aggregates in vessels. There is no evidence of vessel wall involvement (i.e., “vasculitis”).

A B Figure 6. A, Photograph of fundus of the eye in a patient with systemic lupus erythematosus and visual blurring. B, Fluorescein angiograrn of the same eye, showing intraretinal hemorrhages and nonperfusion of a segment of the retinal vein. Arrows indicate vascular occlusion, which is apparent on fluorescein angiogram, while not discernible on ordinary routine photograph.

638 HOPKINS ET AL

Normal Pre- SLE Pregnant Eclampsia Preonant

Figure 7. Levels of C3a and total immunoreactive C3 in blood samples obtained simultaneously from normal pregnant women, women with preeclampsia, and pregnant women with systemic lupus erythematosus (SLE). Plasma C3a levels were normal, with 1 exception each, in normal subjects during pregnancy, and in women with preeclampsia. The mean C3 level was significantly lower in pregnant SLE patients compared with that in normal pregnant controls (P < 0.05). The plasma C3a value was elevated in 5 of the 7 pregnant SLE patients. The C3a level was unavailable in 1 patient with preeclampsia. Bars show group means; broken lines show normal mean + 2 SD; shaded areas show abnormal values.

Four of these women developed proteinuria during pregnancy, and 1 of the 4 developed hypertension. The total C3 content was normal in 6 of the 7 patients; the mean level for the group was 110 mg/dl, which al- though it was not below the conventional normal range, was significantly lower than that observed in normal pregnant women (P < 0.05). In contrast, plasma C3a levels in 5 of the 7 SLE patients (including all 4 with proteinuria and/or hypertension) were ele- vated above the normal range, with a group mean of 3 10 ng/ml (Figure 7).

The total C3 content was compared with the C3a content in 20 samples drawn simultaneously from the 7 lupus patients who were followed serially during pregnancy. Decreases in the total C3 levels correlated inversely with elevations in the plasma C3a levels (r = - 0.59) (Figure 8). Thus, relative hypocomplemente- mia in pregnant women who have lupus is associated with an increase in the plasma content of complement

split products, indicating that complement activation is occurring.

DISCUSSION We and others (8-10) have previously reported

that complement cleavage products are elevated in the plasma of patients with active SLE. The findings reported here support the notion that elevations of plasma C3a levels predict disease flares and, in preg- nant SLE patients, signal complement activation. We found increased levels of complement cleavage prod- ucts at times when both C3 and C4 were at normal levels. This finding should prove useful in the manage- ment of SLE, since conventional measures of comple- ment activation have been of limited value as predic- tors of disease activity (14-21). Valentijn et a1 (14), for example, noted that the level of CH50 was normal in 35% and the C3 value normal in 53% of patients with clinically active SLE; those observations are consist- ent with the data presented in Table 1 in this report.

600

500 E

.Z 40C ?

30C

\ D

0

Q) W 0 5 20c

1 oc

0 0

0

I I I I 100 150 200 250

C3 ( m g / d l )

Figure 8. Levels of C3a and C3 in 20 blood samples obtained simultaneously from 7 lupus patients who were followed serially during pregnancy. Decreases in the total C3 contents correlated inversely with elevations in the plasma levels of C3a (r = - 0.59).

C3a AND C5a IN SLE 639

The lack of a strong correlation between clinical disease activity and hypocomplementemia has several possible explanations. First, serum levels of C3 and C4 reflect both synthesis and degradation of the compo- nent proteins; these levels may therefore remain within the normal range, despite accelerated consump- tion, should synthesis be increased as part of an acute-phase response to the inflammatory process. Conversely, levels may be persistently low in patients with decreased synthesis or with hereditary deficien- cies of individual complement components (22). The measurement of cleavage products in plasma, there- fore, should provide a more sensitive index of com- plement activation in both the normocomplementemic and hypocomplementemic SLE patient.

The cause of hypocomplementemia in lupus patients who develop proteinuria and/or hypertension during pregnancy is poorly understood. In these pa- tients, decreases in total C3 have been observed while levels of Cls-C1 inhibitor complex have remained normal (23). This has led to a suggestion that the decrease in the C3 level may be the result of decreased complement synthesis (23). However, our studies demonstrate that the decrease in total C3 observed during pregnancy in lupus patients is accompanied by a rise in the cleavage product C3a, and this indicates that accelerated complement activation has occurred. The presence of a normal Cls-C1 inhibitor level in similar patients described by Lockshin et a1 (23) sug- gests that complement activation in this situation may not be occurring via the classical pathway. Whether activation occurs via the alternative pathway or as a result of accelerated intravascular coagulation (i.e., activation of complement by products of the clotting cascade) remains to be determined.

The findings from these studies indicate that the serial measurement of complement split products in plasma will prove useful in the assessment of disease activity. However, we note that C3a and C5a are biologically active products (24-28) that can activate platelets and neutrophils, respectively. Indeed, we have reported that neutrophils derived from patients with active SLE display increased numbers of iC3b receptors (CR3) on their surfaces (29); this finding yielded experimental support for the hypothesis that neutrophils have been activated by circulating stimuli (e.g., C5a). Activated neutrophils (i.e., increased CR3- bearing cells) have also been reported to be present in the peripheral blood of patients undergoing hemo- dialysis or patients with acute thermal injury, syn-

dromes that are characterized by an elevation of complement-derived peptides in plasma (30,3 1).

Complement activation products in plasma, by virtue of their capacity of activate inflammatory cells, may therefore contribute to vascular injury. Analysis of our data as displayed in Figure 3 indicates that this mechanism may be operative, especially in a subset of patients with acute dysfunction of the central nervous system. Specimens obtained postmortem from these patients have characteristically demonstrated an oc- clusive vasculopathy (in the absence of typical vascu- litis), the pathogenesis of which remains unclear (32,33). Moreover, a characteristic vasoocclusive reti- nopathy, similar to that illustrated in Figure 6, has recently been described by Jabs et al (34) in patients with SLE. Analysis of the systemic features in those patients revealed that only CNS disease correlated with the development of this form of retinopathy (34). A similar retinal vasoocclusive disease has been de- scribed in patients with Purtscher’s retinopathy. This syndrome, originally described in patients who had experienced severe (non-ocular) trauma, has recently been observed in patients with pancreatitis and ele- vated plasma levels of C5a (35). It has been suggested that Purtscher’s retinopathy is due to complement- induced retinal leukoembolization (35).

Careful examination of pathologic specimens from 2 of our patients with cerebritis revealed occlu- sion of the small vessels by leukoaggregates (Figures 4 and 5). These findings not only resemble those of the classic Shwartzman reaction (36), but also those ob- served clinically in endotoxemia (the adult respiratory distress syndrome), in cerebral injury following extra- corporeal circulation, and in acute thermal injury. Vascular injury in these conditions has been attributed to the activation of inflammatory cells by complement- derived peptides generated in plasma (12,37,38).

We hypothesize that the pathogenesis of the occlusive vasculopathy observed in both the retina and the central nervous system in SLE patients results from the exposure of platelets and neutrophils in the microvasculature to circulating anaphylatoxins. This hypothesis must account for the absence of clinically apparent CNS disease in all patients with elevations in the levels of plasma anaphylatoxins. First, there may be individuals whose disease is clinically silent. In patients undergoing dialysis, neutropenia and hypox- emia due to complement split products in plasma are generally not noticed by the clinician. Second, the activation of inflammatory cells requires exposure to a threshold concentration of stimulus. Thus, the peak

levels of anaphylatoxins achieved in plasma, as well as the duration of exposure, may be important determi- nants. Finally, it must be noted that in the prototype Shwartzman phenomenon, the presence of endotoxin in plasma is necessary but not sufficient to produce the classic lesion: The skin must first be “prepared” by an intradermal injection of endotoxin administered 4-18 hours prior to the intravenous infusion.

Although Gregory Shwartzman described the phenomenon 50 years ago (36), the requirement of the preparatory injection has remained a puzzle. Recent work has demonstrated that endotoxin as well as endog- enous mediators, such as interleukin-1 (IL-I) and tumor necrosis factor (TNF), induce cultured human vascular endothelium to increase the adhesiveness of their surface for polymorphonuclear leukocytes (39). This process requires the synthesis of new protein by the endothelial cell and may account for the role of the “preparatory” injection in the Shwartzman lesion. Thus, it is possible that the irreversible occlusive vascular lesions of CNS lupus similarly require not only the activation of circu- lating neutrophils and platelets, but also the presence of a receptive vascular endothelium. Candidates responsi- ble for altering vascular endothelium in SLE are numer- ous and would include not only IL-1 and TNF, but also complement activation products, immune complexes, and anticardiolipin antibody.

In summary, this study demonstrates that ele- vations of plasma C3a levels may occur prior to other clinical and laboratory evidence of SLE disease flare. Moreover, the increased generation of C3a in the plasma of pregnant SLE patients implicates comple- ment activation as the cause of hypocomplementemia. Finally, the circulation of complement-derived inflam- matory mediators may represent an unrecognized cause of vascular injury in SLE, particularly in pa- tients with CNS disease.

REFERENCES Fauci TY, Haynes BF, Katz P: The spectrum of vascu- litis. Ann Intern Med 89:660-676, 1978 Lange K, Wasserman E, Slobody LB: Significance of serum complement levels for diagnosis and prognosis of acute and subacute glomerulonephritis and lupus erythe- matosus disseminatus. Ann Intern Med 53:636-646, 1960 Schur PH, Sandson J: Immunological factors and clini- cal activity in systemic lupus erythematosus. N Engl J Med 278533338, I968 Ruddy S, Carpenter CB, Chin KW, Knostman JN, Soter NA, Gotze 0, Miiller-Eberhard HJ, Austen KF: Human complement metabolism: an analysis of 144 studies. Medicine (Baltimore) 54:165-178, 1975

5 .

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

HOPKINS ET AL

Alper CA, Rosen F: Studies of the in vitro behavior of human C’3 in normal subjects and patients. J Clin Invest

Sliwinski AJ, Zvaifler NJ: Decreased synthesis of the third component (C3) in hypocomplementemic systemic lupus erythematosus. Clin Exp Immunol 11:21-29, 1972 Charlesworth JA, Williams DG, Sherington E, Lach- mann PJ, Peters DK: Metabolic studies of the third component of complement and the glycine rich glyco- protein in patients with hypocomplementemia. J Clin Invest 53:1578-1587, 1974 Belmont HM, Hopkins P, Edelson HS, Kaplan HB, Ludewig R, Weissmann G, Abramson S: Complement activation during systemic lupus erythematosus: C3a and C5a anaphylatoxins circulate during exacerbations of disease. Arthritis Rheum 29: 1085-1089, 1986 Falk RJ, Dalmasso AP, Kim Y, Lam S, Michael A: Radioimmunoassay of the attack complex of comple- ment in serum from patients with systemic lupus erythe- matosus. N Engl .I Med 312:1594-1599, 1985 Kerr LD, Adelsberg BR, Spiera H: The correlation between SLE and disease activity, complement compo- nents, and complement activation products (abstract). Arthritis Rheum (suppl) 29(4):S60, 1986 Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, Schaller JG, Tala1 N, Winchester RJ: The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 25: 1271-1277, 1982 Chenoweth DE, Cooper SW, Hugli TE, Stewart RW, Blackstone EH, Kirklin SW: Complement activation during cardiopulmonary bypass. N Engl J Med 304:497- 505, 1981 Mancinis G, Cabibara AB, Hereman JR: Immunochem- ical quantitation of antigens by single radial immunodif- fusion. Immunochemistry 2:235-254, 1965 Valentijn RM, van Overhagen H, Hazevoet HM, Her- mans J, Cats A, Daha MR, van Es LA: The value of complement and immune complex determinations in monitoring disease activity in patients with systemic lupus erythematosus. Arthritis Rheum 28:904-913, 1985 Singsen BH, Bernstein BH, King KK, Hanson V: Sys- temic lupus erythematosus in childhood: correlations between changes in disease activity and serum comple- ment levels. J Pediatr 89:358-369, 1976 Can0 PO, Jerry LM, Sladowski JP, Osterlan CH: Cir- culating immune complexes in systemic lupus erythema- tosus. Clin Exp Immunol 29:197-204, 1977 Abrass CK, Nies KM, Louie JS, Border WA, Glassock RJ: Correlation and predictive accuracy of circulating immune complexes with disease activity in patients with systemic lupus erythematosus. Arthritis Rheum 23:273- 282, 1980 Cairns SA, London A, Mallick NP: The value of three immune complex assays in the management of systemic lupus erythematosus: an assessment of immune complex levels, size, and immunochemical properties in relation to disease activity and manifestation. Clin Exp Immunol

Inman RD, Fong JKK, Pussell BA, Ryan PJ, Hughes GRV: The Clq binding assay in systemic lupus erythe-

46:2021-2034, 1967

40:273-282, 1980

C3a AND C5a IN SLE 641

20.

21.

22.

23.

24.

25.

26.

27,

28.

29.

30.

matosus: discordance with disease activity. Arthritis Rheum 23:1282-1286, 1980 Morrow WJW, Isenberg DA, Todd-Pokropek A, Parry HF, Snaith ML: Useful laboratory measurements in the management of systemic lupus erythematosus. Q J Med

Tron F, Letarte J, Roque-Antunes MC, Lesarve P: Specific detection of circulating DNA:anti-DNA im- mune complexes in human systemic lupus erythema- tosus sera using murine monoclonal anti-DNA antibody. Clin Exp Immunol 49:481-487, 1982 Hunsicker MC, Ruddy S, Carpenter CB, Schur PH, Merrill JR, Miiller-Eberhard HJ, Austen KF: Metabo- lism of the third complement component (C3) in ne- phritis. N Engl J Med 287:835-840, 1972 Lockshin MD, Harpel PC, Cruzin ML, Becker CG, Klein RF, Watson RM, Elkon KB, Reinitz E: Lupus pregnancy. 11. Unusual pattern of hypocomplementemia and thrombocytopenia in the pregnant patient. Arthritis Rheum 2858-66, 1985 Showell HJ, Glovsky MM, Ward PA: C3a induced lysosomal enzyme secretion from human neutrophils. Int Arch Allergy Appl Immunol 67:227-232, 1980 Polley MJ, Nachman RL: Human platelet activation by C3a and C3a des arg. J Exp Med 158:603-615, 1983 Fernandez HN, Henson P, Otani A, Hugli TE: Chemo- tactic response to human C3a and C5a anaphylatoxins. J Immunol 120:109-115, 1978 Bjork J, Hugh TE, Smedegard G: Microvascular effects of anaphylatoxins C3a and C5a. J Immunol 134:1115- 1119, 1985 Van Epps D, Chenoweth DE: Analysis of the binding of flourescent C5a and C3a to human peripheral blood leukocytes. J Immunol 132:2862-2866, 1984 Buyon J, Shadick N, Berkman R, Hopkins P, Dalton J, Weissmann G, Winchester R, Abramson SB: Surface expression of CR3 on neutrophils as a marker of disease activity in systemic lupus erythematosus. Clin Immunol Immunopathol (in press) Arnaout MA, Hakim RM, Todd RF, Dana N , Colten

202:125-138, 1982

HR: Increased expression of an adhesion-promoting surface glycoprotein in the granulocytopenia of hemo- dialysis. N Engl J Med 312:457-462, 1985

31. Moore RD, Davis C, Rodrick M, Mannik J, Fearon D: Neutrophil activation in thermal injury as assessed by increased expression of complement expression. N Engl J Med 314:949-952, 1986

32. Ellis SG, Verity MA: Central nervous system involve- ment in systemic lupus erythematosus: a review of neuropathologic findings in 57 cases, 1955-1977. Semin Arthritis Rheum 8:212-233, 1979

33. Johnson RT, Richardson EP: The neurological manifes- tations of systemic lupus erythematosus: a clinical pathological study of 24 cases and a review of the literature. Medicine (Baltimore) 47:337-369, 1968

34. Jabs DA, Stuart SL, Hochberg MC, Newman SA, Heiner MD, Stevens MB: Severe retinal vaso-occlusive disease in systemic lupus erythematosus. Arch Ophthal- mol 104:558-563, 1986

35. Jacob HS, Goldstein I, Shapiro I, Craddock P, Ham- merschmidt D, Weissmann G: Sudden blindness in acute pancreatitis of complement-induced retinal leukoembo- lization. Arch Intern Med 141:134-136, 1981

36. Shwartzman G: Phenomenon of Local Tissue Reactiv- ity. New York, Paul Hoeber, 1937

37. Jacob HS, Craddock PR, Hammerschmidt D, Moldow LF: Complement-induced granulocyte aggregation: an unsuspected mechanism of disease. N Engl J Med

38. Hammerschmidt D, Weaver L , Craddock P, Jacob H: Association of complement activation and elevated plasma C5a with adult respiratory distress syndrome. Lancet 1:947-949, 1980

39. Pohlman TH, Stanness KA, Beatty PG, Ochs HD, Harlan JM: An endothelial cell surface factor(s) induced in vitro by lipopolysaccharide, interleukin 1 , and tumor necrosis factor-a’ increases neutrophil adherence by a CDwl8-dependent mechanism. J Immunol 136:4548- 4553, 1986

3021789-794, 1980