Evidence-based guidelines for the use of immunologic tests: Anticentromere, Scl-70, and nucleolar antibodies


  • John D. Reveille,

    Corresponding author
    1. University of Texas Health Science Center, Houston
    • Division of Rheumatology and Clinical Immunogenetics, The University of Texas at Houston Health Science Center, 6431 Fannin, MSB 5.270, Houston, TX, 77030
    Search for more papers by this author
  • Daniel H. Solomon,

    1. Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • The American College of Rheumatology Ad Hoc Committee on Immunologic Testing Guidelines

    Search for more papers by this author
    • Members of the American College of Rheumatology Ad Hoc Committee on Immunologic Testing Guidelines are as follows: John D. Reveille, MD: University of Texas Health Science Center, Houston; Daniel H. Solomon, MD, MPH, Peter Schur, MD: Brigham and Women's Hospital, Boston, Massachusetts; Arthur Kavanaugh, MD (chair): University of California at San Diego; Yvonne R. S. Sherrer, MD: Center for Rheumatology, Immunology, and Arthritis, Fort Lauderdale, Florida; Robert Lahita, MD, PhD: St. Vincent's Medical Center, New York, New York.

      The American College of Rheumatology is an independent professional, medical, and scientific society that does not guarantee, warrant, or endorse any commercial product or service.


The autoantibodies classically associated with systemic sclerosis (SSc; scleroderma) include anticentromere antibodies (ACA) and anti–Scl-70 (otherwise known as anti–topoisomerase I or anti–topo I). In addition to these is the less-commonly occurring antinucleolar antibody system, which comprises a mutually exclusive heterogeneous group of autoantibodies that produce nucleolar staining by immunofluorescence (IIF) on cells from a variety of species (1). The most widely recognized of these include anti–PM-Scl (2), anti–U3-RNP (closely crossreactive with antifibrillarin) (3), anti–Th/To (4), and the anti–RNA polymerase family (RNAP), including anti–RNA polymerase I (5), II (6), and III, the latter three not always demonstrating a nucleolar staining pattern (7).

The purpose of the present review is to analyze the published literature on the clinical usefulness of ordering tests of scleroderma-associated autoantibodies in patients with known or suspected scleroderma. The evolution of scleroderma antibody testing will be presented, the different technologies discussed, and the analyzable published data on SSc antibody testing in disease diagnosis presented. Also, the application of SSc antibody testing in prognosis will be presented, specifically in predicting the extent of cutaneous and pulmonary involvement (the literature on other manifestations of SSc was either too sparse or the definition of organ involvement too varied to allow standardization of results and hence formulation of guidelines). Finally, the data regarding longitudinal followup of SSc antibodies will be presented to determine whether there is a role for serial testing in clinical care of the patient with SSc. Other autoantibodies occasionally seen in patients with SSc, including anti–U1-RNP and anti–Ro, are covered in other reports from this committee (R. Lahita et al, manuscript in preparation; J. Reveille et al, manuscript submitted).

Critical review for articles on immunologic laboratory tests

This article is part of a series on immunologic testing guidelines. The introduction to the series outlines the full methodology for obtaining data, grading the literature, combining the information from multiple sources, and for developing recommendations. The introduction was published in Arthritis Care & Research, Volume 47, Number 4, August 15, 2002, pp 429–433.

The published articles reviewed for this article included those identified from electronic databases, Medline, using the following search terms: “scleroderma autoantibodies”; “anti-Scl70 autoantibodies”; “anti-topoisomerase I autoantibodies”; “anti-centromere autoantibodies”; “antinucleolar autoantibodies”; “anti-fibrillarin autoantibodies”; “anti-U3-RNP autoantibodies”; “anti-RNA polymerase autoantibodies”; “anti-Th or To autoantibodies”; and “anti-PM-Scl autoantibodies.” Only English-language articles were considered, and review articles, editorials, individual case reports, case series with less than 10 patients, animal studies, and papers that did not report primary data were excluded from analysis, as were articles that did not report complete data, i.e., showing data only from “representative patients.” To avoid duplication of data, the most inclusive article was selected from the same center or group of investigators. Articles published through December 2000 were included.

All articles were critically reviewed using published standards for studies of diagnostic tests (8–12). Test use was categorized as primarily diagnostic or prognostic (which also included monitoring). Information was extracted from each paper to allow for calculation of a weighted average for sensitivity and specificity; likelihood ratios (LRs) were then derived from these values (positive LR = sensitivity/[1-specificity]; negative LR = [1-sensitivity]/specificity). Recommendations for use of tests were based on the LRs, where a test was considered to be “very useful” for a given disease if the weighted average positive LR was >5 or negative LR was <0.2. A test was considered “useful” if the weighted average positive LR was >2 and ≤5 or negative LR was >0.2 and ≤0.5. A test was considered “not useful” if the positive LR was ≤2 or the negative LR was >0.5.



Since the early 1960s it has been recognized that antinuclear antibodies (ANA) are common in the sera of patients with systemic sclerosis (13, 14). In 1979, a basic, heat-labile, chromatin-associated, nonhistone 70-kd protein antigenic to antibodies from SSc patients, was described, which was isolated from rat liver nuclei using a combination of biologic and immunologic methods. This was initially designated Scl-70 (15). Subsequent analyses revealed this autoantibody system to be directed against topoisomerase I (16, 17).

The following year another ANA associated with systemic sclerosis, ACA, was described (18), where HEp-2 cells were used as the substrate for the ANA. ACA previously had not been seen using IIF on tissue substrates such as mouse liver, because the tissues in question undergo cell division much less commonly. To date, more than 6 centromere proteins are bound by sera from patients with SSc, designated CENP-A (centromeric nucleoprotein A) through CENP-F. Molecular analyses have shown that CENP-A is a 17-kd centromere-specific histone H3-like protein (19). CENP-B is an 80-kd alphoid DNA binding protein (20–22). CENP-C is a 140-kd chromosomal component required for kinetochore assembly (22, 23). CENP-E is a 312-kd kinesin-like motor protein (24). CENP-F is a nuclear matrix protein that accumulates in the nuclear matrix during S phase, assembling onto kinetochores at late G2 during mitosis (24, 25). The fact that the antinucleolar antibody (ANoA) staining pattern of ANA was associated with systemic sclerosis has been known since at least 1970 (26). The first of the ANoA to be characterized in 1977, anti–PM-Scl antibodies, was originally described in patients with myositis-scleroderma overlap syndromes using immunodiffusion (ID) techniques (27). With the refinement of immunoprecipitation techniques, a number of other ANoA specificities have been recognized in the sera of patients with SSc, including anti–Th/To-RNP in 1983 (4, 28), anti–U3-RNP (closely crossreactive with antifibrillarin) in 1985 (29), anti–RNAP I in 1987 (5), and anti–RNAP II and III in 1993 (7, 30) (although it should be noted that the latter do not always show an ANoA pattern on IIF).


Anticentromere antibodies.

Anticentromere antibodies have been most typically determined by their characteristic staining pattern on immunofluorescence on HEp-2 cells (18). Patients with SSc produce autoantibodies recognized by immunoblotting (IB), which react against at least 6 different centromeric proteins, CENP-A–CENP-F (31). However, these distinctions have not been shown to have clinical relevance. All sera containing ACA react with CENP-B (32, 33). A solid-phase enzyme-linked immunosorbent assay (ELISA) has been established using a cloned fusion protein of CENP-B as antigen (33). More recently this ELISA has been refined and has been found to have adequate sensitivity and specificity for clinical use (33–35).

Anti–Scl-70 (anti-topo I).

Anti–Scl-70 antibodies have classically been determined by double ID techniques against calf or rabbit thymus extract, including Ouchterlony and counterimmunoelectrophoresis (15, 18). The Scl-70 antigen was found to be a basic, nonhistone chromosomal protein of 70,000 molecular weight found in rat liver, calf or rabbit thymus, or in HEp-2 or lymphoid cells (18). The antigen is heat-labile, not inactivated by deoxyribonuclease or ribonuclease, also inactivated by proteolytic enzymes such as trypsin or pronase. However, ascertainment of anti–Scl-70 antibodies by ID usually requires 2–3 days, and thus is cumbersome and time consuming, and more difficult to perform in high volume, such as for commercial laboratories. Thus, IB and ELISA more recently have been introduced (16, 17, 36–38). Topoisomerase I, purified from calf thymus glands, initially was used as antigen (38). More recent studies have utilized recombinant topo I fusion proteins as the substrate for the ELISAs (35, 39, 40).

Antinucleolar antibodies.

Anti–PM-Scl was the first ANoA to be characterized, using Ouchterlony double ID tested against reference sera and PM-Scl antigen from rabbit thymus extract (27, 41, 42). Other studies of PM-Scl antibodies as well as analyses of other nucleolar autoantibodies have utilized immunoprecipitation assays using [32P] orthophosphate-labeled or [35S] methionine-labeled cell extracts, including anti–Th/To ribonucleoprotein, anti–U3-RNP antifibrillarin, and anti–PM-Scl (4, 5, 43).

Summary recommendations.

ACA has been traditionally determined by its particular staining pattern on IIF against HEp-2 cells. More recent techniques, including ELISAs and immunoblotting, have been shown to perform with satisfactory accuracy when compared with IIF. Likewise, the traditional technique of ID to determine anti–Scl-70 (anti–topo I) autoantibodies is being replaced in many commercial laboratories by ELISAs. Although initially lacking in specificity, recent reports have suggested ELISA to perform satisfactorily compared with ID. Antinucleolar antibodies and other anti-RNA polymerase antibodies are most commonly determined by immunoprecipitation, a technique not widely commercially available, although ID is available for anti–PM-Scl.

Indications for clinical use of ACA, anti–Scl-70, and ANoA tests

Diagnostic value of ACA.

From the initial literature search, data were extracted and combined from 6 articles encompassing 7 research centers (44–49) to compare SSc and healthy controls. The ACA had a sensitivity of 33% and a specificity of 99.9% (positive likelihood ratio [LR] of 327 and negative LR of 0.7). Ten articles examining the prevalence of ACA in patients with SSc versus patients with other rheumatic diseases (those with primary Raynaud's phenomenon were graded separately) were graded “A” (8–12), met criteria for inclusion (Table 1) (44–47, 50–54), and formed the basis of the calculations of the overall sensitivity recommendations. The overall sensitivity of ACA for diagnosis comparing SSc patients with those with other connective tissue diseases (CTD) was 31% and specificity was 97.4% (positive LR of 12.5 and negative LR of 0.7). To compare ACA in SSc patients versus primary Raynaud's phenomenon, 8 articles were analyzed (47, 48, 50–53, 55, 56). The ACA had a sensitivity of 24.1% and a specificity of 90% (positive LR of 2.3 and negative LR of 0.8). Two studies (57, 58) examined ACA in SSc patients versus non-SSc relatives. The ACA were found in 15 of 79 of the SSc patients versus 1 of 253 of the family members (sensitivity, 19%; specificity 99.6%, positive LR 48, negative LR 0.8). Of note was 1 study from Japan (54) in which the frequency of ACA in patients with other CTD exceeded that of other ethnic groups. The reasons for this are not entirely clear, as the same was not seen at another Japanese site (47).

Table 1. Population statistics for ACA (by IIF) in diagnosis: SSc versus other CTDs and normal controls*
LocationPatientsControlsSensitivity, %Specificity, %Pos LRNeg LR
  • *

    ACA = anticentromere antibodies; IIF = indirect immunofluorescence; SSc = systemic sclerosis; CTDs = connective tissue diseases; Pos = positive; LR = likelihood ratio; Neg = negative.

  • Excluding primary Raynaud's phenomenon.

SSc versus normals      
 Farmington, CT (44)54/1960/8528100>330.7
 Rochester, MN (45)34/820/7142100>510.6
 Paris, France (46)48/820/3059100>440.4
 Melbourne, Australia (49)35/680/4052100>430.5
 Tokyo, Japan (47)78/2640/5230100>220.7
 Tao Yuan, Taiwan (48)11/561/5002099.8980.8
 Bangkok, Thailand (49)1/490/442100>0.91.0
SSc versus other CTDs      
 Pittsburgh, PA (18)14/451/1633199.4520.7
 Farmington, CT (44)54/1962/902897.8130.7
 Rochester, MN (45)34/820/12142100>860.6
 Multicenter, USA (50)1/460/3332100>7.41.0
 London, UK (51)22/751/952998.9270.7
 Paris, France (46)48/820/4859100>690.4
 Groningen, Netherlands (52)7/401/2418964.20.9
 Pisa, Italy (53)32/1513/2402198.8180.8
 Tokyo, Japan (47)78/2647/2483097.2150.7
 Nagoya, Japan (54)49/10125/1514983.43.10.6
SSc versus primary Raynaud's phenomenon      
 Farmington, CT (55)26/8918/12328851.90.8
 Mexico City, Mexico (56)2/233/289600.21.3
 Multicenter, USA (50)1/460/312100>21.0
 London, UK (51)22/751/1529934.40.8
 Bath, UK (61)11/361/1531934.60.7
 Groningen, Netherlands (52)7/400/3018100>6.60.8
 Pisa, Italy (53)32/1514/4821922.60.9
 Tokyo, Japan (47)78/2640/2630100>110.7
 Tao Yuan, Taiwan (48)11/568/2720700.71.1
SSc versus non-SSc relatives      
 London, UK (71)11/580/21819100>510.8
 Tokyo, Japan (58)4/211/3519976.60.8

The presence of ACA has been associated with the CREST (calcinosis, Raynaud's esophageal dysmotility, sclerodactyly, telangectasia) variant of systemic sclerosis (18, 59), a syndrome now thought to represent a subset of limited systemic sclerosis. Five studies comparing the frequency of ACA in patients meeting at least 2 of the 5 criteria for CREST compared with controls were included in the present analysis (44–46, 60, 61) (Table 2). The sensitivity (65%) and specificity (99.9%) were excellent, with positive LR as high as 650 and negative LR of 0.4. Twelve studies compared the frequency of ACA in patients with CREST to other patients with SSc (18, 45, 48, 51, 52, 55, 56, 59, 60, 62–64). The ACA had 61% sensitivity, 84% specificity, positive LR of 3.9, and negative LR of 0.5 for CREST compared with those with SSc. The specificity rose dramatically (98%), as did the positive LR (30) when the frequency of ACA in patients with CREST was compared with that found in other connective tissue disease patients (45, 46, 48, 51, 52, 55, 60). When comparing patients with CREST and primary Raynaud's phenomenon, the specificity and positive LR were somewhat lower (83% and 3.5, respectively) although the negative LR was excellent (0.2).

Table 2. Population statistics for ACA (by IIF) in diagnosis: CREST versus systemic sclerosis, normal controls, and other diseases*
LocationPatientsControlsSensitivity, %Specificity, %Pos LRNeg LR
  • *

    CREST = calcinosis, Raynaud's phenomenon, esophageal dysmotility; sclerodactyly, telangiectasias. For other abbreviations, see Table 1.

  • Excluding primary Raynaud's phenomenon.

CREST versus normals      
 Farmington, CT (44)31/730/8542100>330.6
 Rochester, MN (45)13/190/7168100>510.3
 Paris, France (46)39/450/3087100>2020.1
 Calgary, Alberta (60)26/270/2796100>7280.04
 Tao Yuan, Taiwan (48)11/201/5005599.82750.5
CREST versus SSc      
 Philadelphia, PA (59)15/309/4250792.30.6
 Farmington, CT (55)30/5423/8955742.10.6
 Denver, CO (18)12/212/2457927.10.5
 London, UK (51)20/462/2943936.10.6
 Padova, Italy (62)12/183/1467793.20.4
 Bologna, Italy (63)13/142/709397310.07
 Calgary, Alberta (60)26/273/2696898.40.04
 Rochester, MN (45)13/1921/6368672.10.5
 Mexico City, Mexico (56)5/82/2363917.00.4
 Ioannina, Greece (64)3/82/2638924.90.7
 Groningen, Netherlands (52)7/107/4070834.00.4
 Tao Yuan, Taiwan (48)11/200/3655100>450.5
CREST versus other CTDs      
 Farmington, CT (55)31/732/904398190.6
 Rochester, MN (45)13/190/12168100>2640.3
 Calgary, Alberta (60)26/272/1139698540.04
 London, UK (51)20/461/954399390.6
 Paris, France (46)39/450/4887100>3190.1
 Groningen, Neth. (52)7/101/247096170.3
 Tao Yuan, Taiwan (48)11/2030/1,2315598230.5
CREST versus primary Raynaud's phenomenon      
 Farmington, CT (55)30/5418/12355853.80.5
 Calgary, Alberta (60)26/2718/5896693.10.06
 London, UK (51)20/461/1529934.40.8
 Groningen, Neth. (53)7/100/3018100>6.60.8
 Tao Yuan, Taiwan (48)11/208/2720700.71.1
 Mexico City, Mex. (56)5/83/286394110.4

More recently, an ELISA using a cloned fusion protein of CENP-B as antigen was shown to have no added sensitivity in the diagnosis of CREST versus SSc, primary Raynaud's phenomenon, and other CTDs. Initially, it was found to perform with considerably less specificity than IIF (34), although more recent refinements have improved its performance (35, 36).

Studies examining ethnic differences in the frequency of SSc-associated autoantibodies have found ACA to occur more frequently in whites than Africans (65), African Americans (66, 67), Hispanics (66), or Asians (49, 67).

Summary recommendations.

ACA (determined by IIF) are rarely found in patients with CTDs other than SSc, family members of SSc patients, or healthy controls, and thus are very useful in the diagnosis of SSc, particularly in the CREST variant thereof. In fact, ACA are even useful in distinguishing CREST patients from other patients with SSc or primary Raynaud's phenomenon. At this time there are insufficient data to recommend ACA as determined by techniques other than IIF (ELISA, etc.) in the diagnosis of SSc, and this is one area that needs further investigation.

Diagnostic value of anti–Scl-70 (anti-topo I).

Thirty two studies assessing the prevalence of anti–Scl-70 antibodies in patients with SSc compared with other reference groups were reviewed, of which 17 formed the basis of these recommendations (18, 36, 49–51, 55–58, 61, 68–75) (Table 3). Four studies utilized ID to study anti–Scl-70 antibody prevalence in SSc patients versus “normals” (36, 55, 68, 72). Despite a sensitivity of 20.2% in the pooled analysis, anti–Scl-70 had a 100% specificity for SSc and a positive LR of greater than 25 (negative LR 0.8). Ten studies (18, 36, 50, 51, 55, 57, 58, 68–70) examined anti–Scl-70 autoantibodies by ID in SSc patients versus patients with other connective tissue diseases (other than primary Raynaud's phenomenon). Although the overall sensitivity of anti–Scl-70 for SSc was not high (26%), the specificity was (99.5%), with positive LRs (when compared with other CTD patients) of 52 overall, and >15 in all of the studies. Seven studies examined the presence of anti–Scl-70, as determined by ID, in SSc patients versus those with primary Raynaud's phenomenon (18, 50, 55–57, 61, 68), with an overall sensitivity of 28%, specificity 98%, and positive LR of 10 (negative LR 0.7). Two studies examined anti–Scl-70 by ID in family members of SSc patients (72, 73). Here the specificity and LRs were quite high (sensitivity 25%, specificity 100%, positive LR >83, negative LR 0.8).

Table 3. Population statistics for anti–Scl-70 (anti-topo I) in diagnosis*
 PatientsControlsSensitivity, %Specificity, %Pos LRNeg LR
  • *

    Anti–Scl-70 (ascertained by immunodiffusion unless specified). For abbreviations, see Table 1.

  • Anti–Scl-70 ascertained by immunoblotting.

  • Anti–Scl-70 ascertained by enzyme-linked immunosorbent assay.

SSc versus normals*      
 Farmington, CT (55)18/890/3420100>90.8
 Paris, France (36)6/640/3017100>6.40.8
 Tokyo, Japan (72)14/210/2067100>170.7
 Calgary, Alberta (68)6/260/2723100>8.40.8
SSc versus other CTDs*      
 Pittsburgh, PA (18)9/501/1631899.4300.8
 Farmington, CT (55)18/890/5820100>150.8
 Calgary, Alberta (68)6/260/11323100>340.8
 Multicenter, USA (50)9/462/3332099.4330.8
 London, UK (51)16/751/952198.9190.8
 London, UK (69)157/7350/9321100>250.8
 Paris, France (36)6/640/8117100>170.8
 Warsaw, Poland (70)60/1072/905697.8250.4
 Ioannina, Greece (58)4/120/10033100>510.7
 Pisa, Italy (57)60/1511/2404099.6990.6
SSc versus 1° Raynaud's phenomenon*      
 Pittsburgh, PA (18)9/501/1018901.80.8
 Farmington, CT (55)18/891/1582899.4470.7
 Calgary, Alberta (68)6/260/1423100>3.50.8
 Multicenter, USA (50)9/460/3120100>80.8
 Bath, UK (61)8/361/1522933.30.8
 Pisa, Italy (57)60/1480/4841100>330.6
 Mexico City, Mexico (56)9/234/284077>40.8
SSc versus non-SSc relatives*      
 London, UK (73)6/580/21810.3100>250.9
 Tokyo, Japan (72)14/210/3567100>170.3
SSc versus normals      
 Farmington, CT (55)25/891/3428979.30.7
 Melbourne, Australia (49)18/680/4027100>150.7
 Paris, France (36)31/640/3048100>290.5
 Nijmegen, Netherlands (74)11/260/5442100>400.6
 Bangkok, Thailand (49)37/490/4476100>1390.2
SSc versus other CTDs      
 Paris, France (36)31/648/8148904.80.6
 Nijmegen, Netherlands (74)11/262/3024299.3600.6
 Farmington, CT (75)28/832/583497110.7
SSc versus normals      
 Taipei, Taiwan (71)24/320/2075100>630.3
 Farmington, CT (75)26/830/5031100>510.7
SSc versus other CTDs      
 Taipei, Taiwan (71)24/329/3875763.20.3
 Farmington, CT (75)26/831/583198160.7

IB has also been used in recent studies to determine anti–Scl-70. Four studies (36, 49, 55, 74) examined anti-topo I by IB in SSc patients compared with healthy controls. The sensitivity for SSc was 41%, and specificity and positive LR remained high (99.4% and 68, respectively). Three studies were found comparing the frequency of anti–Scl-70 determined by IB in SSc patients compared with those with other CTD (excluding primary Raynaud's phenomenon) (36, 74, 75). The sensitivity (40%) remained higher than ID, with similar specificity (99%) and positive LR (40).

Despite the widespread commercial use of anti–Scl-70 ELISA, only 2 studies were found satisfying inclusion criteria for this analysis (71, 75), and these were more than 10 years old. The combined sensitivity of anti–Scl-70 for the diagnosis of SSc was slightly higher than IB (43%), with a high specificity and LR compared with “normals” (100% and >55, respectively), although less compared with patients with other CTD (90%, 4.3). Because of the small number and age of these studies and more recent statements that anti–Scl-70 ELISAs performed by commercial laboratories were satisfactory (35, 40), we will not make any recommendations or criticisms regarding anti–Scl-70 ELISAs in this report. Similarly, despite the suggestion that the ELISA was more sensitive than other techniques in detecting anti–Scl-70 antibodies (75), 1 recent review concluded that there was no significant difference in the various methodologies used in the determination of anti–Scl-70 autoantibodies on the sensitivity and specificity for SSc (76). ACA and anti–Scl-70 autoantibodies rarely coexist in the same individuals (77–79). In 670 patients from 9 articles reporting in which both autoantibodies were examined only 3 had both occurring simultaneously (74, 79).

Summary recommendations

Anti–Scl-70 antibodies (determined by ID and IB) are rarely found in patients with connective tissue diseases other than SSc, family members of SSc patients, or in healthy controls, and thus are very useful in the diagnosis of SSc. There are limited published studies providing numbers to calculate likelihood ratios for anti–Scl-70 as determined by ELISA, and those that exist suggest somewhat lower specificity and positive likelihood ratios. Hence it is premature to make any recommendations regarding anti–Scl-70 as determined by ELISA in the diagnosis of SSc or other CTD at this time. More studies are needed in this area.

Diagnostic value of antinucleolar antibodies.

Antinucleolar antibodies have been reported in 15–40% of patients with SSc (41, 42). The number of published reports on frequencies in non-SSc patients and healthy controls meeting criteria for inclusion in this analysis were relatively few, hence comparative analyses between series were limited. Nevertheless, specific ANoA have not been reported in healthy controls (1, 27). Likewise, specific ANoA have not been found to occur in a large series of healthy relatives of SSc patients (80).

The frequency of anti–PM-Scl varies among different ethnic groups, ranging from approximately 3% of SSc patients to 8% of myositis sera (1, 27) in whites to being absent in a large series of 275 Japanese patients with newly diagnosed SSc (81). Three studies examined RNAP antibodies in the diagnosis of SSc (5, 69, 82). Despite a high specificity for the disease for RNAP I and III, as well as for antifibrillarin (closely crossreactive with anti–U3-RNP) and anti–Th/To (83), the infrequency of these antibodies in SSc patient populations limits their predictive value in diagnosis. RNAP II, on the other hand, has also been reported to occur in patients with systemic lupus erythematosus and overlap syndromes (84).

Summary recommendations.

The low sensitivity of ANoA for SSc, despite their high specificity (other than anti–PM-Scl and RNAP II), limits their utility in the diagnosis of SSc; hence it is inappropriate to recommend their usage in the diagnosis of SSc.


From the literature search, 59 articles assessing the correlation between SSc-associated autoantibodies and specific aspects of prognosis met the criteria for further review. Of these, 37 unduplicated series were chosen, graded “A” (8–12), and form the basis of the recommendations (Tables 4–7 (1–3, 7, 34, 36, 41, 42, 51, 53, 59, 63, 64, 70, 80–82, 85–102). It would have been desirable to include other outcomes in this analysis, such as death, disability, or renal failure. However, too few articles that met criteria for inclusion covered these parameters, therefore meaningful statistical analysis was precluded.

Table 4. ACA and limited cutaneous involvement in SSc*
Site (reference)Sensitivity (%)Specificity (%)Pos LRNeg LR
  • *

    ACA ascertained by indirect immunofluorescence unless specified. ACR = American College of Rheumatology; CENP = centromeric nucleoprotein; ELISA = enzyme-linked immunosorbent assay. For other abbreviations, see Table 1.

ACR classification limited cutaneous scleroderma (100)    
 Pittsburgh, PA (41)83/191 (43)203/206 (99)4.30.6
 Philadelphia, PA (59)21/46 (46)23/26 (89)4.20.6
 Chicago, IL (34)27/52 (52)34/44 (77)2.30.6
 Denver, CO (42)19/42 (45)32/34 (94)7.50.6
 Houston, TX (87)39/81 (48)165/188 (88)4.00.6
 Columbia, MO (82)20/53 (38)30/36 (83)6.30.7
 London, UK (51)20/46 (43)20/22 (93)6.10.6
 Bath, UK (99)37/101 (37)21/21 (100)>370.6
 Naples, Italy (89)14/49 (29)40/41 (98)150.7
 Milan, Italy (63)13/14 (93)68/70 (97)310.07
 Nijmegen, Netherlands (95)7/8 (88)24/25 (96)220.1
 Turku, Finland (90)4/26 (15)5/5 (100)>1.10.9
 Adelaide, South Australia (91)26/38 (67)11/11 (100)>280.3
 Overall330/747 (44)676/729 (93)6.10.6
Barnett classification I (85)    
 Baltimore, MD (104)15/24 (63)67/71 (94)10.50.4
 Pisa, Italy (53)22/68 (32)69/83 (93)4.60.7
 Turin, Italy (92)20/31 (65)50/60 (83)3.80.4
 Naples, Italy (93)12/28 (43)66/69 (96)110.6
 Bologna, Italy (94)7/26 (27)9/9 (100)>3.70.7
 Copenhagen, Denmark (96)4/28 (14)127/202 (63)0.41.4
 Melbourne, Australia (85)32/45 (71)23/28 (82)3.90.4
 Tokyo, Japan (98)6/17 (35)41/48 (85)2.30.7
 Overall118/267 (44)452/570 (79)2.10.7
Anti-CENP Antibodies by ELISA (ACR classification for skin involvement (100)    
 Chicago, IL (34)25/52 (48)(89)4.40.6
Table 5. Diagnostic and prognostic associations with antinucleolar antibodies*
Site (reference)TechniqueSensitivity (%)Specificity (%)Pos LRNeg LR
  • *

    Pos = positive; LR = liklihood ratio; Neg = negative; IP = immunoprecipitation; IB = immunoblotting; SSc = systemic sclerosis; ID = immunodiffusion.

  • American College of Rheumatology criteria—diffuse cutaneous scleroderma (100).

  • Barnett classifications II and III (85).

  • §

    Only RNA polymerase III was studied.

Diffuse cutaneous involvement and antifibrillarin (or U3-RNP) autoantibodies     
 Pittsburgh, PA (1)IP13/71 (18)56/65 (86)1.30.9
 Copenhagen, Denmark (96)IP8/202 (4)28/28 (100)>1.21.0
 London, UK (80)IP3/58 (5)34/35 (97)1.61.0
 Tokyo, Japan (81)IP5/71 (7)200/204 (98)3.50.9
 Houston, TX (87)IB, IP40/188 (13)239/242 (96)3.30.9
 Overall 69/590 (12)557/574 (97)4.00.9
Diffuse cutaneous involvement and RNA polymerase autoantibodies     
 Columbia, MO (82)IP8/36 (22)52/53 (98)110.8
 London, UK (101)IP32/69 (46)52/61 (85)3.10.6
 Bath, UK (99)IP14/21 (67)85/101 (84)4.20.4
 Tokyo, Japan (81)IP13/70 (19)203/204 (99.6)480.8
 Pittsburgh, PA (7)§IB, IP50/111 (45)107/114 (94)7.50.6
 Overall 117/307 (38)499/533 (94)60.7
PM-Scl and SSc/myositis overlap syndromes (versus SSc w/o myositis)     
 Pittsburgh, PA (2)ID10/41 (24)353/359 (98)120.8
 Warsaw, Poland (3)ID24/27 (89)265/269 (98.5)590.1
 Overall 34/68 (50)618/628 (98)310.5
Table 6. Anti–Scl-70 antibodies and diffuse cutaneous involvement in SSc*
Site (reference)Sensitivity (%)Specificity (%)Pos LRNeg LR
  • *

    ACR = American College of Rheumatology; ELISA = enzyme-linked immunosorbent assay. For other abbreviations, see Table 1.

  • Anti–Scl-70 determined from a commercial ELISA.

  • Anti–Scl-70 determined from ELISA.

  • §

    Anti–Scl-70 determined from immunoprecipitation.

  • Anti–Scl-70 determined from immunoblotting.

ACR criteria—diffuse cutaneous scleroderma (100): immunodiffusion    
 Pittsburgh, PA (41)68/206 (33)157/191 (82)1.80.8
 Philadelphia, PA (59)10/26 (38)44/46 (96)9.50.6
 Baltimore, MD (86)4/22 (18)25/25 (100)>7.40.8
 Columbia, MO (82)9/36 (25)49/53 (92)3.10.8
 Houston, TX (87)40/188 (21)77/81 (95)4.20.8
 London, UK (51)10/29 (35)40/46 (87)2.70.7
 London, UK (88)7/38 (18)34/37 (92)2.30.8
 Warsaw, Poland (70)30/39 (77)38/68 (56)1.80.4
 Ioannina, Greece (64)15/26 (58)6/8 (75)2.30.6
 Naples, Italy (89)28/41 (68)21/49 (43)1.20.7
 Milan, Italy (63)42/70 (60)14/14 (100)>230.4
 Overall263/721 (37)505/618 (82)2.00.8
ACR criteria—diffuse cutaneous scleroderma (100): other techniques    
 Turku, Finland (90)4/5 (80)18/26 (85)5.30.2
 Chicago, IL (34)11/44 (25)48/52 (92)3.10.8
 Bath, UK (99)§9/21 (43)77/101 (76)1.80.8
 Nijmegen, Netherlands (95)11/25 (44)9/9 (100)>80.6
 Tokyo, Japan (81)41/71 (58)91/112 (81)3.10.5
 Overall76/166 (46)243/300 (81)2.40.7
Barnett classifications II and III (80): immunodiffusion    
 Pisa, Italy (53)48/83 (58)55/68 (79)2.80.5
 Turin, Italy (92)24/59 (75)31/31 (100)>220.3
 Naples, Italy (93)56/69 (81)17/28 (61)2.10.3
 Bologna, Italy (94)6/9 (67)21/26 (81)3.50.4
 Copenhagen, Denmark (96)29/202 (14)26/28 (93)2.00.9
 Melbourne, Australia (85)10/29 (35)41/45 (91)3.90.7
 Tokyo, Japan (98)28/48 (58)15/17 (88)4.80.5
 Overall201/499 (40)206/243 (85)2.70.7
Barnett classifications II and III (85): ELISA    
 Tel-Hashomer, Israel (97)70/151 (46)18/18 (100)>160.6
Table 7. Use of ACA and Anti–Scl-70 antibodies in predicting radiographic pulmonary fibrosis among patients with SSc*
Site (reference)Sensitivity (%)Specificity (%)Pos LRNeg LR
  • *

    Anticentromere antibodies (ACA) tested by indirect immunofluorescence (IIF) against HEp-2 cells. SSc = systemic sclerosis; Pos = positive; LR = liklihood ratio; Neg = negative; ID = immunodiffusion.

  • Done by immunoprecipitation.

  • Done by enzyme-linked immunosorbent assay.

  • §

    Done by immunoblotting.

ACA (by IIF)    
 Pittsburgh, PA (41)29/175 (34)193/222 (87)2.60.8
 Philadelphia, PA (59)0/15 (0)33/57 (50)<0.06n.a.
 Houston, TX (87)3/68 (4)125/178 (74)0.21.2
 London, UK (51)7/36 (19)24/39 (62)0.501.3
 Bath, UK (99)9/54 (17)33/74 (45)0.31.9
 Turku, Finland (90)2/14 (17)15/17 (88)1.30.9
 Tokyo, Japan (81)9/142 (6)97/132 (74)0.241.3
 Naples, Italy (93)7/57 (12)37/48 (77)0.551.1
 Bologna, Italy (94)1/11 (9)18/24 (75)0.361.2
 Copenhagen, Denmark (96)9/37 (24)123/193 (64)0.671.2
 Nijmegen, Netherlands (95)1/9 (11)18/25 (72)0.391.2
 Tokyo, Japan (98)2/40 (5)14/25 (64)0.141.5
 Overall80/658 (12)730/1,034 (71)0.411.2
Anti–Scl-70 (by ID)    
 Pittsburgh, PA (41)58/175 (57)164/222 (74)2.20.6
 Philadelphia, PA (59)3/15 (20)48/57 (85)1.51.0
 Houston, TX (87)20/68 (29)156/178 (88)2.40.8
 London, UK (51)12/35 (34)31/40 (78)>240.8
 London, UK (88)10/42 (24)33/33 (100)2.60.9
 Naples, Italy (93)41/57 (72)19/48 (40)1.20.7
 Bologna, Italy (94)5/11 (45)18/24 (75)1.80.7
 Copenhagen, Denmark (96)15/37 (41)176/193 (92)4.90.6
 Tokyo, Japan (98)23/40 (58)18/25 (72)2.10.6
 Overall182/408 (45)663/820 (81)2.30.7
Anti–Scl-70 (by other techniques)    
 Bath, UK (99)23/54 (43)64/74 (86)3.10.7
 Turku, Finland (90)5/14 (42)10/17 (59)1.01.0
 Tokyo, Japan (81)§59/144 (41)112/121 (93)5.50.6
 Nijmegen, Netherlands (95)§4/9 (44)18/25 (72)1.60.8
 Paris, France (36)§18/34 (53)17/30 (57)1.20.8
 Overall109/253 (43)221/267 (83)2.50.7

Cutaneous involvement in SSc has been quantitated by 2 different measures. In criteria established by the American College of Rheumatology (ACR) limited cutaneous scleroderma (lcSSc) is defined as involvement distal to the elbows and knees (103), and diffuse cutaneous scleroderma (dcSSc) as involvement proximal to the elbows. In the criteria of Barnett et al (85), 3 levels of cutaneous involvement have been established. In type I disease, involvement is restricted distal to the metacarpophalangeal joints (MCPs). Type II disease includes arm, leg, and face involvement, and type III disease includes truncal involvement. Twenty-one studies were reviewed examining ACA for defining the extent of cutaneous disease (34, 41, 42, 51, 53, 59, 63, 82, 85, 87, 89–96, 98, 99, 104) (Table 4). The sensitivity of ACA in predicting the presence of lcSSC using ACR criteria (96) was 44%; specificity 93%; and positive LR of 6.1. In 8 studies (53, 85, 92–94, 96, 98, 104) including 267 patients with type I involvement by Barnett's criteria (85), the sensitivity for limited skin involvement was similar (47%), but both the specificity and positive LR were considerably less robust (81% and 2.5, respectively). This was due to the relatively high frequency of ACA seen in patients with Barnett class II disease in many of the studies (data not shown). These data suggest that ACA perform best in predicting limited cutaneous involvement in SSc distal to the elbows and knees, as opposed to distal to the MCPs. Some of the patients deemed as having “lcSSc” were early in their disease course, and may have ultimately developed dcSSc.

Three studies assessing anti–PM-Scl autoantibodies and cutaneous involvement in SSc met criteria for inclusion (80–82) (data not shown). Although highly specific for lcSSc (100%), the low sensitivity (13%) and relatively low LRs (>3) compared with ACA limits the utility of PM-Scl autoantibodies in predicting limited cutaneous involvement. Only 2 studies examined anti–Th/To in predicting cutaneous involvement in SSc, 1 using ACR criteria (80) and 1 using Barnett's criteria (95). Again, the sensitivity for limited skin involvement was low (13%), specificity 96%, positive LR 3.3 (data not shown).

The association of anti–Scl-70 autoantibodies (as determined by ID) with diffuse cutaneous involvement in SSc (by ACR criteria) has been examined in 11 studies of independent series of patients (40, 51, 59, 63, 64, 70, 81, 82, 86–89, 96, 99, 100) (Table 6). Overall, anti–Scl-70 antibodies had a sensitivity of 37%, and a specificity of 82%, with a relatively low positive LR of 2.0 (Table 6). Only 5 articles were found using techniques other than ID for the determination of anti–Scl-70 (34, 81, 90, 95, 99) to predict the presence of dcSSc. Although the small number of articles and different techniques employed limit any meaningful conclusions, considering the data from the 5 studies together, the other techniques yielded greater sensitivity (46%) and similar specificity (81%) and positive LR (2.4) as ID. Using the Barnett criteria, 7 articles were found examining the correlation of intermediate and diffuse cutaneous involvement (types II and III) with the presence of anti–Scl-70 by ID (53, 85, 92–94, 96, 98). The overall sensitivity was 40%, specificity 85%, and positive LR 2.7. Only 1 study examined a technique other than ID (ELISA) to determine the extent of cutaneous involvement using Barnett's classification (98), and excellent specificity (100%) and positive LR (>16) was found, but it is difficult to draw a meaningful conclusion from this single study.

Five nonoverlapping studies examining cutaneous involvement in SSc and antifibrillarin nucleolar antibodies (or U3-RNP) (1, 80, 81, 87, 96) and 5 RNA polymerase antibodies (7, 81, 82, 99, 100) met criteria for inclusion in the present analysis (Table 5). Although both autoantibodies had a high specificity for diffuse cutaneous involvement (96% and 93%, respectively), antifibrillarin or anti–U3RNP antibodies were markedly less sensitive (12% versus 36%) and had lower predictive value (4 versus 6) compared with anti-RNAP antibodies, which had the highest positive LR for diffuse skin involvement among the other SSc-associated autoantibodies.

The absence of ACA and the presence of anti–Scl-70 both have been associated with radiographic interstitial pulmonary fibrosis per se (41, 51, 78, 81, 88, 105) or for the severity thereof (102). Pulmonary involvement in SSc has been defined by numerous measures, most commonly either by radiographic interstitial fibrosis or by abnormalities on pulmonary function tests (PFTs), most utilizing the forced vital capacity (FVC) or the diffusion capacity for carbon monoxide (DLCO). In total, 14 series were identified examining correlations of ACA and/or anti–Scl-70 antibodies with radiographic pulmonary fibrosis (36, 41, 51, 59, 81, 87, 88, 90, 93, 94, 96, 98, 99, 102) (Table 7). Two of these examined only anti–Scl-70 autoantibodies (36, 88). The presence of ACA by IIF against HEp-2 cells was associated with a lower likelihood of radiographic pulmonary fibrosis (sensitivity 12%, specificity 71%, positive LR 0.4) (Table 5). On the other hand, the presence of anti–Scl-70, determined by ID, was positively associated with radiographic pulmonary fibrosis (sensitivity 45%, specificity 81%, positive LR 2.3). The number of published series examining other methods of determining anti–Scl-70 autoantibodies that met criteria for inclusion in the present analysis is small (n = 5) and the methods varied: 3 used IB (36, 81, 95), 1 used IP (103), and 1 used ELISA (90). Thus it is impossible to formulate guidelines for a specific technique other than ID at this time. Nevertheless, it is noteworthy that pooling the results of these 5 studies produced similar sensitivity, specificity, and positive LRs as ID (Table 7).

The different criteria employed for defining restrictive lung disease (RLD) (7, 59, 63, 64, 95, 96, 100, 102) make it difficult to compare studies of pulmonary function tests (Table 8). The results of these studies are highly variable. Some showed ACA to be associated with a lower frequency of RLD (63) or with higher FVCs, total lung capacities, and DLCOs (64); and anti–Scl-70 to associated with RLD (7) and with a higher rate of decline of PFTs (102). Others showed no association of RLD with ACA (63, 95) or anti–Scl-70 (59, 63, 95, 96). One study was particularly noteworthy in demonstrating ACA-positive patients to be more likely to have an abnormal DLCO but normal chest radiograph and FVC (96). In another (102), the presence of RNAP II autoantibodies independently predicted lower lung function (even when ethnicity, age, smoking history, and disease duration were simultaneously considered). However, this association of RNAP with pulmonary involvement was not seen in another large series of patients (7). Hence it is impossible to create guidelines regarding the predictive value of SSc-associated antibodies and PFT based on the available literature.

Table 8. ACA and Anti–Scl-70 autoantibodies and pulmonary function test abnormalities in patients with SSc*
Site (reference)Definition of abnormal PFTResult
  • *

    ACA = anticentromere antibodies; SSc = systemic sclerosis; PFT = pulmonary function test; FEV1 = forced expiratory volume 1; FVC = forced vital capacity; DLco = diffusion capacity for carbon monoxide; TLC = total lung capacity; CXR = chest x-ray; CREST = calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias; ELISA = enzyme-linked immunosorbent assay; dcSSc = diffuse cutaneous scleroderma; lcSSc = limited cutaneous scleroderma; CENP = centromeric nucleoprotein.

Baltimore, MD (102)Rate of change of FEV1, FVC, DLcoAfrican Americans and anti–Scl-70 positives had a higher rate of decline in PFTs
Philadelphia, PA (59)Restrictive lung disease = TLC < 80%ACA associated with less restrictive lung disease, anti–Scl-70 not significantly associated with more or less restrictive lung disease
Pittsburgh, PA (7)FVC < 60%2/29 with RNA polymerase III versus 7/21 anti–Scl-70 positives
Copenhagen, Denmark (96)FVC or DLco < 80%51% of ACA positives versus 28% of ACA negative patients have abnormal DLco but normal CXR and FVC (P = 0.005)
Nijmegen, Netherlands (95)Restrictive lung disease = FVC < 80%No association of restrictive lung disease with ACA or Scl-70 antibodies
Ioannina, Greece (64)FVC, TLC < 80%, DLco < 75%ACA positives had a greater, anti–Scl-70 lower FVC, TLC, DLco
Milan, Italy (63)FVC < 80% or FEV1/FVC < 75%No association with either ACA or CREST
Tours, France (100)TLC < 80%, DLco < 75%Anti–Scl-70 (by ELISA) associated with pulmonary involvement in dcSSc patients but not in lcSSc.
Bath, UK (99)DLco < 80%Those with anti-CENP antibodies (determined by ELISA) associated with a lower rate of pulmonary involvement (9/42 versus 59/113; P < 0.005); anti–Scl-70 associated with a higher rate of pulmonary involvement (23/33–69.7% versus 13/122 (32.6%); P < 0.001)

Anti–PM-Scl has classically been associated with the SSc/myositis overlap (106–108). Two studies met criteria and were included in the present analysis (2, 5). Although varying widely in the 2 studies, anti–PM-Scl occurred in 34 of 68 SSc patients with the SSc, myositis overlap overall, compared with 10 of 628 SSc patients without myositis (sensitivity 50%, specificity 98.4%, positive LR 31).

Summary recommendations.

Determination of ACA by IIF early in the course of SSc is very useful in predicting whether skin involvement will remain limited (i.e., distal to the elbows, if not to the MCPs) and is useful in predicting a lesser likelihood for the development of radiographic pulmonary fibrosis (and possibly restrictive lung disease). The presence of anti–Scl-70 antibodies determined by ID is useful in predicting a greater likelihood for the development of diffuse cutaneous involvement, radiographic pulmonary fibrosis, and abnormalities in pulmonary function tests. A limited number of studies examining ACA and anti–Scl-70 by other techniques suggest similar sensitivity, specificity, and LRs, but it is premature to make specific recommendations regarding their use at this time.

Longitudinal assessment.

Seven articles were reviewed that examined serial determinations of ACA and anti–Scl-70 autoantibodies in SSc patients over time (96, 109–114). The different methodologies and study designs employed make it impossible to combine the data, so the results of each are presented individually (Table 9). In 3 studies from 1 institution, ACA-positive SSc patients remained positive in nearly all determinations, by either IIF or IB (109–111) (Table 9). Likewise, in 6 studies of anti–Scl-70 positive patients from 3 different institutions (95, 109, 111–113), patients initially anti–Scl-70 positive and patients initially anti–Scl-70 negative remained thus. No correlation with disease activity could likewise be correlated with anti–Scl-70 or ACA levels as determined by ELISA (111). However, of particular interest was 1 recent study of 8 patients, of whom 6 were within 16 months of disease onset and where serial samples were collected for up to 7 years. Antibody levels (as determined by immunoblotting) varied widely over time (113), suggesting continuous antigen presentation and regulation of the anti–Scl-70 immune response over time. In another study of 28 patients followed with sequential anti–Scl-70 determinations, anti–Scl-70 antibodies disappeared in 6 patients. The 6 SSc patients in whom anti–Scl-70 autoantibodies disappeared had less extensive skin and lung involvement than those in whom they persisted (114).

Table 9. ACA, anti–Scl-70, and antinucleolar antibodies in studies of longitudinal assessment in SSc*
Site (reference)Findings
  • *

    IB = immunoblotting; IIF = indirect immunofluorescence; ID = immunodiffusion. For other abbreviations, see Table 8.

Farmington, CT (109)The presence of ACA at study entry in 77 patients with Raynaud's phenomenon correlated with telangectasia. Of 15 patients positive initially for ACA by IB, 14 were consistently positive on multiple determinations by IIF or IB. Of 70 patients tested for anti-topo by ID, IB, or ELISA, positive results were obtained in 1, 5, and 4, respectively. Three of the 5 IB positives were ELISA negative. The 1 ID positive remained thus. Three of the initially positive 4 ELISAs turned negative.
Farmington, CT (110)In a longitudinal retrospective study of 37 patients previously tested for ACA who had at least 3 sera available over a mean period of 30.7 months, all but 1 remained ACA positive. Of 22 patients who were initially all ACA negative and who had a mean of 4 sera available over a mean period of 33.5 months, all remained negative.
Farmington, CT (111)Of 103 serum samples from 13 ACA positive and 6 anti–Scl-70 positive patients studied over a period of 3–17 years, anti–CENP-B and anti–Scl-70 showed evidence of isotype switch with persistence of IgG expression. Level of expression of different ACA and anti–Scl-70 isotypes not associated with disease expression.
Nijmegen, Netherlands (95)Longitudinal studies of 7 anti–Scl-70 positive and 9 negative patients show no differences in positivity or negativity or in immunoblotting profiles, despite slight variations in blot band intensity.
Frieburg, Germany (112)A retrospective analysis of 13 anti–Scl-70 positive patients who had at least 3 sera stored over a period of at least 2 years. Titers of anti–Scl-70 antibodies as determined by ELISA remained stable in 9 of 13. One patient acquired and 1 patient lost anti–Scl-70 autoantibodies over the duration of the study.
Baltimore, MD (113)Eight patients with SSc known to be anti–Scl-70 positive followed over time. Anti–Scl-70 antibodies in both early disease and late disease serum samples preferentially recognized the Scl-70 molecule rather than recombinant peptides, suggesting preferential recognition of conformational determinants on Scl-70 throughout the disease course.
Tokyo, Japan (114)28 patients followed with sequential anti–Scl-70 determinations, where the 6 SSc patients in whom anti–Scl-70 autoantibodies disappeared had less extensive skin and lung involvement than those in whom they persisted.

Summary recommendations.

Although the small number of articles concerning serial determination of SSc-associated autoantibodies and the varying methodologies employed make formulation of specific guidelines impossible, it is clear that once a patient is determined as being anti–Scl-70 or ACA positive or negative, there is little justification for serial determinations.


We acknowledge the helpful input and comments of Drs. Hank Homburger and Russ Tomar of the College of American Pathologists and of Dr. Marvin Fritzler of the University of Calgary.