"AJRCCM" new method: qualitatively uncertain lung nodules!

In recent years, as people pay more attention to health and improve the resolution and image quality of chest CT examinations, the detection rate of lung nodules and early lung cancer has increased, which has reduced the mortality rate of lung cancer by 26%-61%

Although lung nodules are closely related to lung cancer, benign lesions such as inflammation and granuloma can also appear as nodules

Because there is no histological diagnosis, these nodules cannot be determined as benign or malignant and are called indeterminate lung nodules

In clinical practice, the error rate of IPN diagnosis and the proportion of IPN patients undergoing invasive procedures such as needle biopsy and surgery are relatively high

In order to avoid missed diagnosis and over-diagnosis and treatment, it is very important to seek a non-invasive and accurate diagnosis method

Lung cancer screening and the establishment of lung cancer prediction models are of great significance to the early diagnosis of lung cancer

Among them, PET-CT is one of the most commonly used diagnostic and staging tools for lung cancer, but it is expensive, with a false positive rate of 25% (high incidence of infectious diseases can reach 39%) [2]

Some clinically applied models, such as the Mayo clinical model based on clinical variables such as patient age, smoking history, tumor history in other parts, and nodule characteristics [3]

In the high-risk category, a Mayo score lower than 0.

However, the diagnosis of benign and malignant pulmonary nodules is still inaccurate, especially for patients in the intermediate-risk group

Other non-invasive diagnostic methods, such as blood markers, imaging examinations, etc.

To this end, the team led by Professor Michael N Kammer conducted a prospective collection, retrospective blinding (PRoBE) study, and constructed a new lung cancer prediction model-including three biomarkers (clinical variables, hematology) Biomarker model (CBM) combined with imaging omics) [4]

Among them, the use of CBM can reduce the invasive surgery rate of patients with moderate-risk benign nodules from 62.

This research was recently published in the American Journal of Respiratory and Critical Care Medicine (AJRCCM)

 The study included 4 IPN case-control cohorts from 4 different medical centers: VUMC cohort (n=171), DECAMP cohort (n=99), UPMC cohort (n=99) and UC Denver cohort (n=88).

The Kammer team first used the Mayo clinical model to divide these patients into low-, medium-, and high-risk groups, and most of the IPN (67%) of all patients were classified as medium-risk groups

For the VUMC cohort, the Brooke University model includes emphysema, the number of lung nodules, the solid component of the nodules, and the family history of lung cancer, with additional risk scores [5]
.

At the same time, they also collected the patient's hematology and imaging data and followed up until the patient was diagnosed with lung cancer or benign nodules
.

Among all blood biomarkers of lung cancer, multiple studies have found that the serum highly sensitive cytokeratin 19 fragment (hs-CYFRA 21-1) has a high diagnostic value for IPN [6]
.

This study by the Kammer team also proved that hs-CYFRA 21-1 is more accurate than Mayo clinical model alone, and the combination of hs-CYFRA 21-1 and Mayo clinical model is better than hs-CYFRA 21-1 alone
.
The diagnostic accuracy rate is high
.
 

Next, the Kammer team evaluated the risk of IPN based on CT imaging omics
.

In all four cohorts, the diagnostic accuracy of imaging omics was higher than the Mayo clinical model
.

Combining imaging omics with the Mayo clinical model can improve the diagnostic accuracy of both
.

It can be seen that the combined model is more advantageous than the single model in diagnosis, so the researchers constructed a new lung cancer prediction model CBM
.

CBM includes Mayo clinical model, serum hs-CYFRA 21-1 level and CT imaging omics
.

In order to verify the accuracy of the model, first train the CBM in the VUMC cohort, and calculate the difference in risk assessment after adding other biomarkers to the Mayo clinical model
.

The results show that the CBM diagnostic accuracy than single model Mayo Clinic, Mayo Clinic genomics + model image, the model Mayo Clinic + hs-CYFRA 21-1 levels (p value P <1.
7 × 10 ^-.
7 , P = 2.
196 × 10 ^-5 and p=2.
966×10 ^-3 )
.

Then verify CBM in the remaining three cohorts
.
The Kammer team found that in the four cohorts, the diagnostic accuracy of CBM was higher than the Mayo clinical model
.

In addition, for the subgroup undergoing PET-CT examination, compared with another Herder model that has been used clinically [7], the Mayo clinical model has a higher diagnostic efficiency in the VUMC cohort (AUC=0.
66 vs.
AUC= 0.
71), but the diagnostic performance in the UC Denver cohort decreased (AUC=0.
88 vs.
AUC=0.
83), but in these two groups, CBM had the highest diagnostic performance
.

After verifying the diagnostic accuracy of CBM in an independent cohort, the Kammer team finally merged the four cohorts (n=456), re-adjusted and calibrated the model to improve the applicability of the model in the clinic
.

As expected, the diagnostic accuracy of CBM in the combined cohort was better than the Mayo clinical model, blood biomarkers, and imaging omics.
The Mayo clinical model’s AUC increased by 0.
124 (95%CI: 0.
091- 0.
156, p<2×10 ^-16 )
.

Overall, CBM has improved the accuracy of clinical diagnosis in a series of situations
.
Compared with Mayo clinical model and expensive PET-CT, CBM only needs to increase CT imaging analysis of lung nodules, fast and cheap hematology detection
.

More importantly, compared with the existing prediction models, the combined biomarker model based on the three biomarkers of clinical variables, hematology and imaging omics not only significantly improved the diagnostic accuracy of IPN patients, but also achieved The early diagnosis also reduces unnecessary invasive surgery
.

I look forward to the early promotion of CBM to clinical practice to benefit patients with pulmonary nodules!

[1] de Koning, HJ, CM van der Aalst, PA de Jong, et al.
, Reduced Lung-Cancer Mortality with Volume CT Screening in a Randomized Trial.
N Engl J Med.
2020;382(6):503-513 .
doi:10.
1056/NEJMoa1911793

[2] Deppen, SA, JD Blume, CD Kensinger, et al.
, Accuracy of FDG-PET to diagnose lung cancer in areas with infectious lung disease: a meta-analysis.
Jama.
2014;312(12):1227-1236 .
doi:10.
1001/jama.
2014.
11488

[3] Gould, MK, L.
Ananth, and PG Barnett, A clinical model to estimate the pretest probability of lung cancer in patients with solitary pulmonary nodules.
Chest.
2007;131(2):383-388.
doi:10.
1378/ chest.
06-1261

[4] Kammer, MN, DA Lakhani, AB Balar, et al.
, Integrated Biomarkers for the Management of Indeterminate Pulmonary Nodules.
Am J Respir Crit Care Med.
2021;204(11):1306-1316.
doi:10.
1164/rccm .
202012-4438OC

[5] McWilliams, A.
, MC Tammemagi, JR Mayo, et al.
, Probability of cancer in pulmonary nodules detected on first screening CT.
N Engl J Med.
2013;369(10):910-919.
doi:10.
1056/ NEJMoa1214726

[6] Kammer, MN, AK Kussrow, RL Webster, et al.
, Compensated Interferometry Measures of CYFRA 21-1 Improve Diagnosis of Lung Cancer.
ACS Comb Sci.
2019;21(6):465-472.
doi:10.
1021/ acscombsci.
9b00022

[7] Herder, GJ, H.
van Tinteren, RP Golding, et al.
, Clinical prediction model to characterize pulmonary nodules: validation and added value of 18F-fluorodeoxyglucose positron emission tomography.
Chest.
2005;128(4):2490- 2496.
doi:10.
1378/chest.
128.
4.
2490