Table 2 Previous studies on early keratoconus with Corvis ST

Hwang et al. [48]

US

31.8 ± 13.4

56.6

VAE-NT (AKC in the original text) (30, 30)

No findings

No

CDVA of 20/20

Clinical KC

ART max yielded an AUROC of 0.739 (sensitivity of 56.7% and specificity of 88.3%)

Awad et al. [49]

Egypt

30.6 ± 9.2 (14–44)

–

FFKC (48, 48)

Normal

Normal topography (Pentacam), mean K < 47 D, I-S ≤ 1.4 D

–

Clinical KC

ART (AUROC 0.88) was a highly sensitive parameter in the FFKC cases

Shajari et al. [50]

Germany

Matched control group (32 ± 11)

–

FFKC (normal tomography in the original text) (27, 27)

Normal

Normal topography (according to elevation maps, corneal thickness, maximum K, and D-index)

–

Clinical KC

ARTmax (AUROC 0.613) was not strong in differentiating populations at early stages of keratoconus

Zhang et al. [51]

China

21 ± 9

43.5

FFKC (23, 23)

Normal

Normal topography with no asymmetric bowtie and no focal or inferior steepening pattern

–

Clinical KC

Corneal deflection amplitude during the first applanation, length at the first applanation, corneal deflection amplitude during the second applanation, and maximum deformation amplitude \(\uparrow\) in FFKC

Tian et al. [52]

China

23.6 ± 8.7

–

FFKC (36, 36)

Normal

No abnormal or suspect tomography

–

Clinical KC

A1T, SP-A1, CBI were significantly softer in the FFKC; diagnosis model using backpropagation neural network (AUROC 0.877) was more sensitive in the detection FFKC than the CBI (AUROC 0.610) and TBI (AUROC 0.659)

Chen et al. [53]

China

22.93 ± 4.91 (12–34)

70.3

FFKC (91, 91)

Normal

Corneal tomography was relatively normal (no asymmetrical bowtie type—oblique radial axis and no central or lower area steep), I-S < 1.4 D, KISA% < 60%)

–

Clinical KC

A1 dArc length (AUROC 0.901), highest concavity radius (AUROC 0.879), A2T (AUROC 0.877), and TBI (AUROC 0.874)

Luz et al. [28]

Brazil

25.5 ± 7.2

–

FFKC (21, 21)

Normal

KISA% < 60% (Placido-disk topography) without a suspect pattern

–

Clinical KC

BAD-D alone (AUROC 0.91 ± 0.057), highest AUROC (AUROC 0.953, sensitivity 85.71%, specificity 98.68%) for a logistic regression model by adding BAD-D, ART Max, and thinnest point related elevation on both the front and back surface

Liu et al. [54]

China

22.00 ± 6.26

59.3

VAE (27, 27)

Normal

Normal topography (mean K < 47 D and I-S value ≤ 1.4 D)

CDVA of 20/20, thickness at thinnest point > 470 μm

Clinical KC

The AUROC of the SSI were not as good as TBI (AUROC 0.928), BAD-D (AUROC 0.926) and CBI (AUROC 0.860)

Koc et al. [55]

Turkey

27.7 ± 6.9

47.6

SKC (21, 21)

Normal

Normal topographic, normal topometric and tomographic findings

–

Clinical KC

TBI AUROC (0.790, cut-off 0.29, sensitivity 67%, specificity 86%). Significant differences were found in the values of A2L, A1V, A2V, and TBI

Xian et al. [56]

China

–

–

FFKC (44, 44)

Normal

Topographic (paracentral I-S ≤ 1.4 D), and tomographic (central anterior and posterior elevations < 8 μm and 13 μm, respectively) examinations

With the Best-Fit-Sphere as the reference sphere

Clinical KC

↓ SP-A1 and \(\uparrow\) TBI in FFKC eyes but AUROCs of SP-A1 and TBI were lower than 0.7

Wang et al. [57]

US

33.8 ± 10.6

–

FFKC (21, 21)

Normal

No tomographical signs

–

Clinical KC

CBI (AUROC 0.785, sensitivity 63.2%) and BAD-D (AUROC 0.757, sensitivity 52.6%) with a common specificity of 80.3%

Miao et al. [58]

China

22.76 ± 4.99

75.3

FFKC (194, 194)

Normal

Normal topography (mean K < 47.00 D; KISA% < 60%; I-S < 1.4 D)

–

Clinical KC

AUROC of SSI II was the highest in distinguishing normal eyes from FFKC, followed by ART and CBI

Zhang et al. [59]

Secondary analysis

–

–

–

–

–

–

–

SP-A1 (AUROC 0.87, sensitivity 0.71, and specificity 0.85) was the only Corvis ST output parameter sensitive to FFKC except the CBI

Peris-Martinez et al. [61]

Spain

Men (26 ± 13)

Women (31 ± 19)

61.5

SKC (13, 16)

Normal

Topography normal with no asymmetric bowtie, and no focal or inferior steepening pattern

–

Clinical KC

Significant differences were found in A1T and A2T, maximum deformation amplitude, highest concavity radius, and A2L and A2V

Steinberg et al. [62]

Germany

31 ± 10

–

SKC (NA, 27)

Normal

KISA% index < 60%, I-S < 1.4 D, and Kmax ≤ 47 D

–

Clinical KC

None of the A1L, A2L, radius of the inward-bended cornea, and deflection length at the highest concavity parameters showed differences

Castro-Luna et al. [63]

Spain

40.21 ± 13.19

–

SKC (20, 20)

No slit-lamp findings

Minor topographic signs of keratoconus and suspicious topographic findings (mild asymmetric bowtie, with or without deviation)

Mean K < 46.5 D; minimum corneal thickness > 490 μm;

Clinical KC

SP-A1 and A2T were the most critical determinants. The random forest model was a good model for classifying SKC (specificity 93%, sensitivity 86%)

Ren et al. [64]

China

22.79 ± 5.78

–

SKC (100, 100)

Normal

No clear evidence of KC

–

Clinical KC

SP-A1 (AUROC 0.753) and CBI (AUROC 0.703) showed significant differences between normal and SKC eyes

Heidari et al. [65]

Iran

30.15 ± 5.42

–

SKC (79, 79)

No finding

Abnormal findings in topography and tomography maps

–

Clinical KC

AUROCs for SP-A1 (0.779), ARTh (AUROC 0.718, CBI (AUROC 0.758), and TBI (AUROC 0.828), were all inferior to the Sirius symmetry index of back (0.908) and Pentacam I-S value (0.862)

Chan et al. [66]

China

–

–

SKC (23, 23)

Normal appearing

Atypical or suspect topography findings that did not meet the diagnostic criteria for KC, with average K ≤ 49 D or HOAs ≤ 1.50 µm in either eye or normal topography

–

Clinical KC

Significant differences were found in BAD-D and TBI between SKC and normal. The TBI and BAD-D showed the highest AUROC (0.925 and 0.786)

Song et al. [67]

China

24.87 ± 7.36

52.9

SKC (70, 70)

Normal

I-S asymmetry and/or bowtie pattern with skewed radial axes (with/without)

–

Clinical KC

AUROC of TBI and BAD-D was 0.944 and 0.965, respectively

Augustin et al. [69]

Germany

27.4 ± 9.3

100

VAE-NT (14, 14)

Normal

Tomographically normal eyes (BAD-D < 1.6 and regular ABCD grading system*)

–

Clinical KC

High sensitivity of both CBI (99.1%) and TBI (99.6%) in detecting tomographic abnormal keratoconus

Herber et al. [70]

Italy

–

–

VAE-NT (18, 18), VAE-NTT (55, 55)

Normal

Topographically normal (VAE-NT), topographically and tomographically normal (VAE-NTT)

–

Clinical KC

CBI provided high sensitivity and specificity of 93.1% to distinguish normal eyes from VAE-NT and VAE-NTT using a cut-off value of 0.2

Ambrósio et al. [35]

Brazil, Italy

–

57.4

VAE-NT (94, 94)

Normal

KISA% < 60%, I-S < 1.45 D

–

Clinical KC

The AUROCs for the TBI, BAD-D, and CBI were 0.985, 0.839, and 0.822 in the VAE-NT group. A TBI cut-off value of 0.29 provided 90.4% sensitivity with 96% specificity

Kataria et al. [71]

India

22 ± 10

–

VAE-NT (also written as SKC) (100, 100)

Normal

VAE with normal topography (TMS-4), KISA% < 60%, I-S < 1.45 D

–

Clinical KC

The TBI (AUROC 0.90) was superior to CBI (AUROC 0.78), BAD-D (AUROC 0.81) and SP-A1 (AUROC 0.76). The TBI (with a 0.63 cutoff) showed the highest accuracy (99.5%), with 99% sensitivity, 100% specificity

Wallace et al. [72]

New Zealand

26.2 ± 10.1

67

VAE-NT (21, 21)

Normal

VAE with normal tomography (Pentacam), KISA% < 60%, I-S < 1.45 D

–

Clinical KC

The TBI (AUROC 0.92) was superior to CBI (AUROC 0.78) and BAD-D (AUROC 0.81). At a cutoff of 0.72, TBI has 99% sensitivity, 67% specificity, and 92% accuracy

Salomao et al. [73]

Brazil, Italy

–

–

VAE-NT (NA, 125)

Normal

VAE with relatively normal topography

–

Clinical KC

AUROC of the TBI was 0.966, BAD-D (0.834), CBI (0.774)

Sedaghat et al. [29]

Iran

26.2 ± 4.3

42

ACE-NT (128, 128)

Normal

Normal pattern and index (Pentacam)

–

Clinical KC

TBI has the best accuracy (AUROC 0.966) for differential diagnosis with a cutoff of 0.24

Steinberg et al. [34]

Germany

–

–

VAE-NT (NA, 32)

VAE-NTT (NA, 18)

Normal

VAE with regular topography (VAE-NT) or regular topography and tomography (VAE-NTT)

–

Clinical KC

The accuracy was reproducible (accuracy in current study population with an optimized TBI cut-off: 0.72), to differentiate between normal and VAE-NT in the study population

Ferreira et al. [74]

Brazil

33.26 ± 14.41

59.6

VAE-NT (57, 57)

Normal

Normal topography (KISA% < 60% and I-S < 1.45 D)

–

Clinical KC

In the VAE-NT group, optimized TBI cut-off value of 0.295 provided a sensitivity of 89.5% and a specificity of 91.0% (AUC 0.960; 95% CI 0.937–0.983)

Koh et al. [75]

Japan

47.2 ± 10.2

–

VAE-NT (23, 23)

Normal

Normal topography (TMS-4)

–

Clinical KC

The AUROC for the BAD-D, CBI, and TBI were 0.668, 0.660, and 0.751, respectively. The TBI cut-off of 0.259 provided 52.17% sensitivity and 88.57% specificity. Nine VAE-NT cases (39.1%) exhibited normal values for the BAD-D, CBI, and TBI. 40% of VAE-NT eyes were classified as normal by the BAD-D, CBI, and TBI

Fraenkel et al. [31]

Germany

33.6 ± 13.3

65

VAE-NT (26, 26)

Normal

Tomographically normal part eye in very asymmetrical corneal ectasia

–

Clinical KC

TBI of the VAE-NT (0.19 ± 0.25) did not differ significantly. Five (19.2%) of 26 eyes had a TBI more than 0.29 and were considered pathological. The VAE-NT eyes (8.5 ± 1.5 mm Hg) showed a significantly more pathological CH and CRF (8.3 ± 1.5 mm Hg) compared with the normal eyes

Augustin et al. [76]

Germany

–

–

VAE-NT (34, 34)

Normal

Tomographically regular fellow eyes by Pentacam AXL

–

Clinical KC

The TBI showed slightly higher sensitivity than the CBI (62% vs. 53%) for detecting keratoconus. 21% of the keratoconus partner eyes could not be recognized as conspicuous, either by CBI or TBI

Padmanabhan et al. [77]

Italy, Brazil

34.29 ± 14.28

–

VAE-NT 105, 105)

Normal

Fellow eye with normal topography (Pentacam HR)

–

Clinical KC

No differences in the SSI were observed between healthy individuals and VAE-NT cases