Lung cancer screening in India: Preparing for the future using smart tools & biomarkers to identify highest risk individuals

Workforce requirement for traditional lung cancer screening, by the numbers

Resourcing an LCS requires significant infrastructure and workforce. The scan itself is performed on a standard CT machine. However, the radiation dose is lower at 1.4 millisieverts (mSv) compared with 7mSV for a diagnostic CT of the chest; the LDCT scan is quick and completed in less than 5 min; however, these scans may compete with time slots for patients who need diagnostic scans, straining resources. It is estimated that a lung cancer navigator who works with the primary care physician is necessary to keep track of high-risk individuals who are sent for LCS. On an average, one navigator is required for 1000 high-risk individuals (to provide counselling, shared decision making, referral to a smoking cessation programme, coordinating with a multidisciplinary lung nodule clinic, and follow up of indeterminate nodules as per established guidelines and tracking patients yearly, in addition to facilitating timely referral to cancer services in the event of lung cancer detection). The workforce includes radiologists, radiology technicians, pulmonologists, interventional radiologists, thoracic surgeons, nurse coordinators, radiation and medical oncologists and data managers. The programme requires a dedicated program manager to oversee personnel and maintain quality audits and controls.

(i) Scale

Implementing LCS in India without overcoming local barriers is impractical. Identifying 1-2 regions with high lung cancer and smoking prevalence to pilot test solutions should be part of the initial strategy. The specified population will require in-depth data collection on lung cancer incidence and demographics over the previous five-years. Registry data and GATS2 data for the region can then be combined to identify individuals with incident lung cancer who used tobacco. The scope of this demographic data will define the number of navigators needed to serve the population for tobacco cessation counselling and referral to a comprehensive LCS programme. Currently, primary care physicians in regional hospitals provide initial care and education on non-communicable diseases in general (this includes oral cavity, breast, and cervical cancer). One such innovative model is the Tata Digital Nerve Center (DiNC) model, which seeks to decentralize care, empower the rural primary care centres, and refer back to centralized resources for complex care, including cancer ( https://www.tcs.com/corporate-social-responsibility/empowerment/known-citizen-drive-DiNC ).

Such a model will provide both accurate data on lung cancer prevalence and tobacco use patterns and provide opportunistic screening and primary prevention through a decentralized model.

(ii) Optimizing opportunistic prevention and screening

Opportunistic LCS can occur when an individual presents for a yearly health examination, a screening mammogram, PAP smear or colonoscopy. At the time of these visits, if the individual fulfils the proposed criteria for LCS, they may be referred to an organized LCS programme. Such opportunistic screening identified lung cancers in women presenting for a yearly mammogram^24,25. ‘Opportunistic’ LCS also occurs when individuals present for CT scans for other respiratory or non-respiratory-related indications. In this instance, the results have been met with variable success²⁶. However, such a program can pose significant challenges, particularly if a well-defined clinical pathway that triages the patient from ‘abnormal nodule’ in the incidental chest CT to diagnosis and therapy is absent. Clinical pathways wherein the radiologists alert the primary care provider of an incidental suspicious lung nodule may be useful²⁷. These patients need to be directed to a robust multi-disciplinary lung nodule programme to avoid unnecessary thoracotomy or other invasive procedures and be followed on a dedicated navigator dashboard according to defined criteria, such as the updated Fleischner criteria²⁸ for incidental nodule management.

(iii) Optimizing nodule diagnostic algorithm

One of the challenges in a screening programme is the appropriate approach to screen-detected nodules. Fewer than four per cent of nodules detected in the NLST were malignant, so a mechanism of identifying and tracking indeterminate nodules is paramount to the success of LCS.

Though the lung imaging, reporting and data system (lung-RADS) introduced in 2015 represented an important step in reducing uncertainty in reporting, it still requires a human interface and trained radiologists²⁹. Computer-aided diagnostic techniques can simultaneously improve nodule detection and reduce the number of nodules that require interpretation by a radiologist. In a decentralized model, this can reduce the number of nodules requiring further evaluation, including biopsies³⁰. Another potential application of technology is cloud-based computerized systems, which have reduced the variability across sites in positivity rates³¹. Radiomics has shown promise in lung nodule evaluation, which can be incorporated into the initial studies³².

(iv) Using robust algorithms to identify at-risk individuals

Current LCS trials target individuals who are at the highest risk based on their tobacco use history. For example, the NLST identified individuals between 55-74 yr with at least 30- pack years of smoking history³³. The revised US Preventive Services Task Force (UPSTF) criteria expanded this in 2021 to include a 20-pack-year smoking history for LCS. Even with these revised criteria, just over half of the African American patients would qualify for LCS, whereas 3/4th of white patients would qualify. Simply changing the criteria to numbers of years smoked (>20 yr of smoking) reduced this disparity, such that over 80 per cent of both ethnic groups qualified for LCS³⁴. Similar personalized risk models may be necessary in India to identify patients at risk for lung cancer to reduce disparities among urban, rural, and ethnic groups.

The most popular form of smoked tobacco in India is in the form of bidis, and the traditional definitions of ‘pack-years’ may not be feasible. Other predisposing factors may play a role (e.g., air pollution and exposure to PM³⁵. In the Indian context, indoor air pollution due to cooking stoves using biomass fuels like wood, dung, or crop residues has been associated with the risk of lung cancer in women^36,37. However, there are currently no guidelines to identify populations eligible for LCS other than the widely used (albeit imperfect) metric of the pack-year. A modified algorithm based on the duration and intensity of tobacco smoking, air pollution, and occupational exposure will need to be considered.

(v) Identifying blood-based biomarkers to aid early diagnosis/screening

A robust, validated point-of-care blood test would be most valuable in resource-constrained countries, such as India and other LMICs. However, such tests will need to be cost-effective, have a high positive predictive value, and be rigorously tested prospectively with longitudinal follow-up to demonstrate a reduction in lung cancer-related mortality. Current ‘liquid biopsy’ testing includes the study of circulating tumour cells, cell-free DNA (cfDNA), circulating tumour (ctDNA), miRNA, exosomes and tumour-educated exosomes (TEX). A single blood tube may provide data on several biomarkers at once and allow for a comprehensive risk readout based on genomic, epigenomic and proteomic biomarkers. A complete discussion of the pros and cons of each of these blood biomarkers is beyond the scope of this paper, but this subject is reviewed comprehensively by Freitas et al³⁸. Such blood-based biomarkers may be complementary to the LDCT screen³⁹.

Compared to genomic aberrations involving changes in the sequence of the DNA, epigenetic regulation affects the function of a particular gene⁴⁰. The most well-studied of these epigenetic phenomena is DNA methylation, characterized by the reversible addition of a methyl group (CH3) to cytosine in DNA to form 5-methyl cytosine. Aberrant methylation patterns (hypomethylation or hypermethylation) are a hallmark of carcinogenesis, occurring early and across the genome, resulting in genomic instability. These epigenetic changes are remarkably distinct across organ types, enabling some recent technologies in this space. Currently, several competing platforms are using early epigenetic changes to detect cancers; one of these involves detecting methylation changes across the genome in the form of a multi-cancer detection test that includes lung cancer, reviewed recently by Constantin et al⁴¹. The purported specificity of GRAIL is >95 per cent with a false positive of less than one per cent. Other methylation-based tests of plasma markers like Lung Epi-check have shown promising results with high sensitivity (87.2%) and specificity (64.2%) in diagnosing early lung cancers. They could be a potential biomarker for screening lung cancer⁴². None of these platforms have long-term follow up of positive tests. Therefore, it is safe to say that these platforms require more rigorous and longitudinal testing in various populations before being adopted into clinical practice.

MicroRNAs (miRNAs) usually have 19-22 nucleotides involved in gene regulation and are very stable in circulation. With just 2 miRNAs, lung cancer patients can be differentiated from individuals without cancer⁴³.However, validation studies are required. Studies combining promising miRNAs and LDCT may increase detection rates⁴⁴. The role of autoantibody-based biomarkers for screening is evolving. Early CDT-lung test, an ELISA-based detection of seven autoantibodies in peripheral blood, followed by a CT scan has shown promise⁴⁵; but the sensitivity for early lung cancers was only 21 per cent and may need to be combined with LDCT.

Taken together, several blood-based biomarker platforms in the multi-cancer early detection space are being evaluated across the world in disparate populations. They may serve as a good adjunct to current screening practices and could be incorporated as a study tool to eventually identify the highest-risk individuals who might benefit from radiographic screening.

(vi) Incorporation of ‘Smart Tools’

At each step of the LCS programme, smart tools developed by Artificial Intelligence (AI) can aid in the workflow, as reviewed by Gandhi et al⁴⁶. This could start from risk stratification models that are unique for India, available on a provider-facing interface to screen high-risk individuals, use of AI algorithms such as SYBIL to ascertain the medium-term risk of lung cancer from single baseline LDCT⁴⁷, ascertain genomics of newly diagnosed lung cancers from imaging or initial pathology slides^48,49, and use of advanced radiomic algorithms that can inform prognosis following radiotherapy for early stage lung cancers⁵⁰.

At present, we have little information about the usefulness of lung cancer screening in India. There is only one prospective study by Singh et al⁵. However, this is a single-arm study with only 253 participants. Nevertheless, the results could shed important preliminary information that may inform future longitudinal studies.