TABLE 1.
Tests used to predict thrombosisa
| Reference, patients, and testb | Sample size | Prevc | Positive testd | PPVe | Sense | Spece | Odds ratiof | Signifg |
|---|---|---|---|---|---|---|---|---|
| 35,h DVT/PE patients: aCL > 33 | 90 | 1 | 0.21 | NA | (0.21) | NA | 5.3i | 0.01 |
| 28,j any thrombosis, virtually all SLE patients | ||||||||
| aCL > 65.1 | 117 | 0.21 | 0.10 | 0.75 | NA | NA | 11.6 | Yes |
| aCL 21.4–65.0 | 117 | 0.21 | 0.37 | 0.26 | NA | NA | 1.3 | No |
| aCL < 21.3 | 117 | 0.21 | 0.53 | 0.06 | NA | NA | 0.3 | Yes |
| 1, any thrombosis, SLE patients | ||||||||
| aCL > 24 | 390 | 0.05 | 0.47 | (0.07) | (0.65) | (0.54) | 1.4i | 0.114 |
| aPTT > 35 s | 390 | 0.05 | 0.17 | (0.15) | (0.50) | (0.85) | 3.38i | <0.001 |
| 30,k any thrombosis | ||||||||
| aCL > 40 | 252 | 0.10 | 0.28 | (0.24) | (0.65) | (0.76) | 3.66l | <0.01 |
| Past thrombosis | 360 | 0.09 | 0.325 | (0.21) | (0.74) | (0.72) | 4.90l | <0.005 |
| 5,m PAPS or SAPS vs SLE without APS | ||||||||
| aCL > 5 | 120 | 0.59 | 0.52 | (0.81) | (0.70) | (0.76) | 2.8 | NA |
| Anti-B2GPI | 120 | 0.59 | 0.33 | (0.95) | (0.54) | (0.96) | 13 | NA |
| 81,n any thrombosis: aCL > 20 or LA present | 474 | 0.09 | 0.28 | (0.22) | (0.64) | (0.76) | 2.7 | NA |
| 44,o SLE with any thrombosis vs SLE without thrombosis | ||||||||
| LA | 175 | 0.25 | 0.25 | 0.59 | 0.59 | 0.86 | 4.3 | NA |
| aCL > 10 | 175 | 0.25 | 0.47 | 0.34 | 0.64 | 0.59 | 1.5 | NA |
| Anti-B2GPI | 175 | 0.25 | 0.17 | 0.47 | 0.32 | 0.88 | 2.6 | NA |
| 37,p definite or probable APS vs various others | ||||||||
| aCL > 30 | 280 | 0.08 | 0.23 | 0.31 | 0.87 | 0.83 | 5.1 | NA |
| Anti-B2GPI | 280 | 0.08 | 0.08 | 0.90 | 0.83 | 0.99 | 106 | NA |
| 78q DVT/PE patients | ||||||||
| Recurrence of VTE, aCL > 5 | 412 | 0.16 | 0.17 | 0.29 | (0.30) | (0.86) | 2.1l | Yes |
| Death, aCL > 5 | 412 | 0.07 | 0.17 | 0.15 | (0.33) | (0.85) | 1.8l | No |
| 76, APS vs not APS in SLE patients with a positive aCL: anti-B2GPI | 65 | 0.45 | 0.34 | (0.82) | 0.62 | (0.89) | 5.6 | 0.0001r |
Prev, prevalence; PPV, positive predictive value; Sens, sensitivity; Spec, specificity; Signif, significance; DVT/PE, deep venous thrombosis/pulmonary embolus; NA, not available; VTE, venous thromboembolism, aPTT, activated partial thromboplastin time; PAPS and SAPS, primary and secondary APS, respectively.
The standard to which the test is compared is indicated. For APS this was definite clinical cases unless otherwise stated, for venous thromboembolism this was diagnosis by appropriate tests (i.e., Doppler ultrasound, high-probability ventilation perfusion scan, or in some cases pulmonary angiography), and for arterial thrombosis tests depended on the affected organ. All IgG aCL units are standardized to GPL sera as supplied by the Anti-Phospholipid Standardization Laboratory, unless otherwise indicated. Despite this, the normal range of healthy controls somehow varies tremendously, ranging from <5 U (78) to <24 U (1). The Physicians study (35) used a different reference standard, supplied by M. Lockshin. The 95th and 98th percentiles in that study were at 33 and 38 U, respectively. We suggest that their data in raw units should not be compared to data from other studies. Anti-B2GPI ELISAs are not standardized, so no units are given. Only IgG anti-B2GPI results have been cited in this paper, although IgM and IgA anti-B2GPI results may also be of relevance.
Frequency of patients with the clinical condition.
Frequency of patients with a positive test result.
Data that were not given in the paper or that are of questionable relevance are in parentheses.
Calculated point-specific likelihood ratio (not given in the paper), unless indicated otherwise.
Significance of the odds ratio given in the paper. If a 95% confidence interval is given for a relative risk, a range is said to be significant if it is outside 1.
Frozen serum samples of 90 patients were compared with the frozen sera of 90 age-, sex-, aspirin allocation-, and smoking-matched controls.
Univariate analysis.
The data are stratum-specific likelihood ratios, not point-specific likelihood ratios. Sensitivity and specificity do not fit the stratum concept well and are misleading if they include the higher strata in this context.
Not all patients were tested for aCL. Three hundred twenty-six patients had LA, as an inclusion criterion. No aCL-positive, LA-negative patient had thrombosis. This was a prospective study.
Multivariate analysis.
Some patients in the SLE group had clinical features of APS but were repetitively negative for aPL. Thrombosis had occurred in 87% of the APS patients and 12% of the patients with SLE without APS.
Numbers of patients with IgG aCL > 20 or who were LA positive were not given. The high-positive group had higher levels of preexisting thrombosis, although most patients in the study were being investigated for miscarriage. This was a prospective study.
Multivariate odds ratios are given in the paper but only for venous and arterial thromboses separately. The raw data of these divisions was not given; therefore risk ratios have not been used.
The manufacturer’s preferred cutoff was 30 GPL; however, a cutoff at 15 GPL (mean + 5 SD) gives a positive predictive value of 0.16.
Analysis of only the 6-month treatment subgroup; this was a prospective study.
Chi-square test.