Skip to main content

Table 3 Predictors, methods, and performance of the assessed development models

From: Inconsistencies in predictive models based on exhaled volatile organic compounds for distinguishing between benign pulmonary nodules and lung cancer: a systematic review

Study

Study participants

Included variables

Statistical method

Modeling method

Validation method

Number

Sensitivity (%)

Specificity (%)

Accuracy (%)

AUC

VOCs

no-VOCs

Peled, N. [20]

53 patients with MPN, 19 with BPN

1-octene

NR

Wilcoxon/Kruskal– Wallis test

DFA

Internal validation: LOOCV

1

â‘  86%

â‘  96.00%

â‘  88.00%

â‘  0.986

Bousamra, M., 2nd [21]

107 patients with LC, 40 patients with BPNs

2-butanone; 3-hydroxy-2-butanone; 2-hydroxy-acetaldehyde; 4-hydroxyhexenal

NR

Wilcoxon test  +   ANOVA

NR

NR

9

① NR; ② 21%; ③ 60%; ④ 84%; ⑤ 92%; ⑥ 28%; ⑦ 70%;⑧ 89%; ⑨ 95%

① NR; ② 100%; ③ 94%; ④ 74%; ⑤ 45%; ⑥ 100%; ⑦ 95%; ⑧ 74%; ⑨ 45%

① -⑨ NR

â‘  0.8;

② -⑨ NR

Fu, X. A. [22]

97 patients with MPN, 32 patients with BPN

2-butanone; 2-hydroxyacetaldehyde; 3-hydroxy-2-butanone; 4-hydroxyhexenal

NR

Wilcoxon test

NR

NR

1

â‘  89.80%

â‘  81.30%

â‘  87.60%

â‘  NR

Li, M. X. [61]

85 patients with MPN, 34 patients with BPN

hydroxyacetaldehyde; 2-butanone; 4-hydroxy-2-hexenal; a mixture of 2-pentanone and pentanal; 3-hydroxy-2-butanone

NR

Kruskal– Wallis test

PLS  +   SVM  +   RF  +  LDA  +   QDA

Internal validation: LCOGV

14

â‘  100%; â‘¡ -â‘­ NR

â‘  81.8%; â‘¡ -â‘­ NR

① 0.946; ② 0.892; ③ 0.892; ④ 0.865; ⑤ 0.811; ⑥ -⑭ NR

① -⑤ NR;⑥ 0.901; ⑦ 0.821; ⑧ 0.694; ⑨ 0.793; ⑩ 0.625; ⑪ 0.696; ⑫ 0.793;⑬ 0.66; ⑭ 0.874

Schumer, E. M. [23]

156 patientd with LC, 65 patients with BPNs

2-butanone, 3-hydroxy-2-butanone, 2-hydroxyacetaldehyde, 4-hydroxyhexanal

NR

NR

NR

NR

3

â‘  93.6%; â‘¡ 76.9%; â‘¢ 40.4%

â‘  44.6%; â‘¡ 78.5%; â‘¢ 90.8%

â‘  -â‘¢ NR

â‘  -â‘¢ NR

Phillips, M. [27]

65 patients with MPN, 15 patients with BPN

1,4-butanediol; 2-pentanamine, 4-methyl-; 2-propanamine; 3-butenamide; 4-penten-2-ol; acetamide, 2-cyanoalanine; N-methylglycine; octodrine

NR

NR

fuzzy logistic regression

Blinded validation; Internal validation: cross-validation

2

â‘  80.1%;

â‘¡ NR

â‘  75%;

â‘¡ NR

â‘  80.5%; â‘¡ NR

â‘  0.88;

â‘¡ 0.80

Shaohua Xie [49]

104 with MPN, 43 patients with BPN

Cyclopentane; Pentane, 3-methyl-; ethylbenzene; N, N-di(methyl)formamide

age

t-test +  Mann-Whitney U test + χ 2 test +  Fisher exact probability

bivariate logistic regression

NR

1

â‘  80.80%

â‘  60.50%

â‘  NR

â‘  0.781

Pengqiang Liao [28]

629 patients with LC, 139 patients with BPN

α -Pinene; 2,3,6-trimethyl-Heptane; 2-Methylnaphthalene; Tetrachloroethylene; Naphthalene; Naphthalene, 1-methyl-; Furan, tetrahydro-2,2,5,5-tetramethyl; p-Cresol; 1-Propene-1-thiol; 5-Hepten-2-one,6-methyl-; Undecane; Methylvinylketone; Dimethadione; 3-Pentenoicacid,4-methyl; 4-Methoxyphenol; Propanoicacid,2-methyl-,butylester; N,N-Dimethylacetamide; tert-Butanol; Methylenechloride; Guaiacol; AceticAcid; benzene; PropionicAcid; Oxetane,2,2-dimethyl; Acetone; Butanoicacid; Cyclohexane and 3 unknown compounds

age, emphysema, tumor size, tumor type, spicule sign

—

GA-SVM +  Binary logistic regression

Internal validation:5-fold cross-validation

2

â‘  51.06%;

â‘¡ 78.7%

â‘ 68.29%; â‘¡68.3%

â‘ 54.15%

â‘¡75.56%

â‘ 0.65; â‘¡0.776

Chen, X. [24]

160 patients with MPN, 70 patients with BPN

1,2-Dichloroethane; Benzene; 2,4-Dimethylhexane; Heptane; Di-tert-butylperoxide; Tetrachloroethylene; Heptanal; Propylcyclohexane;2-Nonenal, (2E); Cyclohexane, heptyl; 1,4-Methanoazulen-9-ol, decahydro-1,5,5,8a-tetramethyl-,(1R,3aR,4 S,8aS,9 S);2,4-Hexadiyne; Butyricacid; 2-methylbutyricacid; Cyclononasiloxane, octadecamethyl-(8CI,9CI); Octadecamethylcyclononasiloxane; Benzoicacid,3,5-bis(1,1-dimethylethyl); METHYL10-METHYLUNDECANOATE; p-Terphenyl

NR

NR

logistic regression + ANN

Internal validation:5-fold cross-validation

4

â‘ -â‘£ NR

â‘ -â‘£ NR

â‘ -â‘£ NR

â‘  0.809; â‘¡ 0.799;â‘¢0.756; â‘£ 0.779;

Rai, S. N. [58]

156 patients with LC, 65 patients with benign PNs

Butyraldehyde; Hexanal;2-Heptanone; Octanal; Undecanal; Butyricacid; Aceticacid; Acrolein; 4-hydroxyhexanal; 4-Hydroxynonenal; Cyclopentanone; Dicyclohexylketone

NR

Boot-SVM-RFE

SVM

Internal validation: 5-fold cross-validation

7

①-⑦NR

①-⑦NR

① 76.98;② 76.07;③ 76.25;④ 78.48;⑤ 76.72; ⑥ 77.65;⑦ 77.92

① -⑦NR

Ding, X. [60]

PKUPH Cohort: 49 lung cancer patients and 51 benign pulmonary nodules

CHPKU Cohort: 39 lung cancer and 24 benign nodules

acetaldehyde, 2-hydroxyacetaldehyde, isoprene, pentanal, butyricacid, toluene,2,5-dimethylfuran, cyclohexanone, hexanal, heptanal, acetophenone, propylcyclohexane, octanal, nonanal, decanal, and2, 2-dimethyldecane

NR

t-test +  Fisher’s exact test

RF

External validation

2

â‘  82.1%;

â‘¡ 71.8%

â‘  92.3%;â‘¡ 76.9%

â‘  84.6%;â‘¡ 73.1%

â‘ 0.872; â‘¡0.744

  1. â‘ -â‘­: model 1 to 14; NR: Not report; Forward SDA: Forward stepwise discriminant analysis; LOOCV: Leave-one-out cross-validation; Fisher SDA: Fisher stepwise discriminant analysis; DFA: Discriminant factor analysis; ANOVA: Analysis of variance; ANN: Artificial neural network; Fisher DA: Fisher discriminant analysis; GA-SVM: Genetic algorithm-support vector machine; PLS: Partial least squares; SVM: Support vector machine classification models; RF: Random forest algorithm; LDA: Linear discriminant analysis; QDA: Quadratic discriminant analysis; LGOCV: Leave group out cross-validation