Most focal persistent ground glass nodules (GGNs) do not progress over 10 years. Research suggests that GGNs that do not progress, those that do, and solid lung cancers are fundamentally different diseases, although histologically they seem similar. Surveillance of GGNs to identify those that gradually progress is safe and does not risk losing a window. GGNs with 5 mm solid component or less than 10 mm consolidation (mediastinal and lung windows, respectively, on thin slice CT) are highly curable with resection. The optimal type of resection is unclear; sublobar resection is reasonable but an adequate margin is critically important.
Most ground glass nodules (GGNs) do not progress.
Traditional (ie, solid, spiculated) lung cancers, GGNs that progress, and GGNs that do not progress seem to be different disease entities.
Observation of GGNs that are predominantly ground glass to identify those that gradually progress is safe.
Lung cancers that have a GGN component have distinctly better long-term outcomes than completely solid lung cancers.
The incidence of recurrence is minimal for completely resected GGNs that have a consolidated portion size of less than 10 mm (on lung windows) or a solid portion size of less than 5 mm (on mediastinal windows).
There has been growing awareness that ground glass nodules (GGNs) are a clinically distinct group of pulmonary nodules. Increased use of chest computed tomography (CT) has led to an increase in nodule detection, which will likely continue with broader adoption of lung cancer screening. Management of GGNs is increasingly relevant in clinical care. The lung cancer community have learned much about how GGNs behave. This chapter reviews relevant evidence that informs the clinical management of patients with a GGN.
This chapter focuses on persistent, focal GGNs with distinct borders in middle-aged to older adults. Such GGNs are strongly correlated with the adenocarcinoma spectrum of non-small cell lung cancer (NSCLC). GGNs due to inflammatory processes (which generally are transient and appear as an infiltrate with indistinct borders) and regional, patchy, or extensive ground glass opacities that indicate focal or diffuse pneumonitis or interstitial lung disease are not included.
The nature of ground glass nodules
A GGN, as defined by the Fleischner Society, is a hazy increased opacity of lung, with preservation of bronchial and vascular margins. This radiographic appearance can be caused by any condition that decreases air in the lung parenchyma without total obstruction of the alveoli, such as interstitial thickening (due to fluid, cells, and/or fibrosis), partial collapse of alveoli, increased capillary blood volume, or a combination thereof.
Assessment of GGNs requires attention to technical details of the CT scan (eg, slice thickness). One study found that more than 50% of lesions considered pure GGNs (without a solid component) on 5 mm slices actually had a solid component on 1 mm slices. Recent guidelines recommend a thin-slice CT (1–1.25 mm) for lung imaging because of such discrepancies. Likewise, small differences cannot be reliably identified when comparing CT scans using different techniques (eg, a 5-mm and a 1-mm thickness scan or a diagnostic CT and the CT done with positron emission tomography [PET] for attenuation correction). Furthermore, there is interobserver variability; one representative study noted 36% discordance for size or presence of a solid area (160 nodules, median diameter 12 mm [5–33] ranging from pure GGN to solid).
Different methods of categorizing GGNs have been used. One approach is to classify GGNs as pure (only ground glass on lung windows), heterogeneous (having both a ground glass and a denser area of consolidation on lung windows), and part-solid (having a solid area that is visible on mediastinal windows, all using thin-slice [1 mm] CT). Another method divides GGNs into quartiles by the percentage of the lesion that is consolidated (on lung windows): 0% to 24% (pure ground glass), 25% to 49%, 50% to 74%, and greater than or equal to 75%. Another strategy estimates the ratio of the size of the ground glass component on lung windows to the solid component on mediastinal windows, called the tumor disappearance ratio (TDR), , or the proportion of consolidation (on lung windows) to ground glass, called a consolidation/tumor ratio (CTR). An older classification distinguishes 6 categories based on the density, morphology, and proportion of a solid component.
In this review, mixed GGN refers to lesions that are neither pure ground glass nor completely solid, that is, containing both a ground glass component and either a consolidated (lung windows) or a solid (mediastinal windows) component. To avoid confusion, the authors specify whether size in cited studies refers to the overall lesion, the consolidated, or the solid component. Finally, unless otherwise noted, the current definition of T-stage categories (based on consolidated or invasive component only) is used instead of the older definition based on overall size.
Pathology of Ground Glass Nodules
The lung adenocarcinoma spectrum begins with premalignant lesions including atypical adenomatous hyperplasia (AAH) and adenocarcinoma in situ (AIS). AAH is a localized, small (≤5 mm) proliferation of mildly to moderately atypical type II pneumocytes. AIS has a predominantly lepidic pattern of neoplastic cells growing along alveolar walls without stromal, vascular, or pleural invasion and less than or equal to 3 cm overall. Minimally invasive adenocarcinoma (MIA) has less than or equal to 5 mm of invasion (≤3 cm overall) in a lepidic background. The diagnosis of AIS or MIA cannot be established from a small biopsy; the entire tumor must be available for histologic examination. Invasive lung adenocarcinomas are typically mixed, and although lepidic predominant adenocarcinoma is most common in GGNs, any subtype can be present and even predominant (lepidic, acinar, papillary, micropapillary, and solid).
The degree of correlation between the radiographic and histologic appearance is not reliable. Resected pure GGNs harbor invasive cancer in 10% to 40% of cases, and 20% to 60% of mixed GGNs are not invasive cancer ( Fig. 1 ). The inclusion of other imaging characteristics (eg, solid size, margin, mass, density) have not identified a reliable way to predict histologic subtype. , Another large study found no correlation between the amount of ground glass on imaging and the lepidic proportion among 465 pIA mixed GGNs (average CTR 0.68).
CT-guided biopsy of GGNs provides reasonable results in several studies, involving either pure or mixed GGNs with a CTR less than or equal to 0.5 and an average overall lesion size of ∼15 mm. In general, about 10% of attempts were nondiagnostic; excluding these, the reported sensitivity is 67% to 95%, and the false-negative rate for a diagnosis of malignancy is 20% to 40%. Sensitivity was only slightly better in larger or more consolidated lesions. Identification of adenocarcinoma subtype or invasiveness was lower, but it is inappropriate to attempt to make such diagnoses on a limited biopsy.
However, a persistent, focal GGN can be predicted to be a form of adenocarcinoma with a high degree of reliability. Adenocarcinoma (of some subtype along the spectrum) is found in ∼95% of mixed GGNs (CTR <0.5) and ∼85% of pure GGNs in multiple studies; these often included cases that underwent immediate biopsy/resection, so the rate of adenocarcinoma in GGNs that have been demonstrated to be persistent is likely even higher.
Natural History of Ground Glass Nodules
Multiple clinical observational studies have found that the majority (60%–90%) of pure or mixed GGNs do not progress during periods of observation for 5 to 10 years. , , , A long-term prospective study found progression in 19% of 243 GGNs (accrued 2000–2005, followed until 2015; progression was defined as increased consolidation on lung windows or growth, usually ≥2 mm). A multicenter prospective study of 1253 GGNs (median 7.8 mm overall size, 85% pure, 7% heterogeneous, 8% part-solid) reported that at 5 years, development of a new solid component (≥2 mm on mediastinal windows) occurred in 6% and 14% of pure and heterogenous GGNs, and growth of a solid component (≥2 mm) was seen in 1%, 5%, and 22% of pure, heterogenous, and part-solid GGNs, respectively. The overall lesion size grew (≥2 mm) in 14% of pure GGNs, 24% of heterogenous, and 48% of part-solid GGNs. A retrospective study of 109 GGNs found that 37% progressed (median observation 4.2 years, Fig. 2 ). , Similar results have been reported by others. ,
These clinical observations confirm multiple basic science observations. GGNs that did or did not progress reveal markedly different genetic patterns (90% harboring epidermal growth factor receptor [ EGFR ] mutations vs 20%, respectively) ; others have corroborated genetic differences. Mouse models demonstrate similar findings: Kirsten rat sarcoma gene ( Kras )-mutated mice develop AAH that does not progress, whereas EGFR -mutated mice develop AAH that progresses to lung adenocarcinoma. These data, together with other studies, suggest varied behavior among premalignant lesions and lung cancers, corresponding to different genetic underpinnings ( Fig. 3 ). ,
Among GGNs that progress, the rate of progression is characteristically very slow. The median time to detection of growth (typically 2–3 mm) was 2 years in one long-term prospective study (range <1–9 years). Similar results were noted in another prospective study: median time to development of a solid component (≥2 mm on mediastinal windows) was 3.8 and 2.1 years for pure and heterogenous GGNs, respectively. Expressed differently, the time until development or growth of a solid component (on mediastinal windows) to greater than or equal to 3.3 mm (a proposed trigger for intervention) was 4.2, 4.2, and 2.5 years for pure, heterogeneous, and part-solid GGNs, respectively; the minimum time to this endpoint was 1.8 and 2.5 years for pure and heterogeneous GGNs and 6 months for part-solid GGNs.
The rate of growth is reported to be ∼2 mm/y among GGNs that grew. In one study 2% were observed to exhibit more rapid growth (>5 mm/y) at least once. Among GGNs that grow, the volume doubling times (VDT) are long (mean 800–1800 days). , , Similar indolent growth rates of GGNs have also been noted in CT screening studies. However, the specific VDT of the solid component (on mediastinal windows) of mixed lesions is reported to be similar to traditional solid lung cancers (100–300 days).
Several studies have investigated predictors of growth by multivariate analysis. , , The most consistent predictors are initial overall lesion size and a history of lung cancer; less consistent predictors are older age and a consolidated/heterogeneous appearance. Few have investigated predictors of progression (increased density, growth of or development of a consolidated or solid component) ; it must be concluded that predictors of progression are insufficiently defined.
In summary, most of the GGNs do not progress, likely reflecting underlying genetic characteristics. The rate of growth among those that grow is slow, with rare exceptions. Predictors of growth include initial overall GGN size and a history of lung cancer.
How should we manage patients?
Can Histologic Features Be Used?
Histology is of limited use in patient management for multiple reasons. Persistent focal GGNs are almost always lesions along the adenocarcinoma spectrum. Bronchoscopic or CT-guided biopsy of small opaque lesions can be difficult, and there is a substantial false-negative rate. The histologic subtype cannot be determined until after resection (ie, after a therapeutic decision has been made). Most importantly, most GGNs do not progress, and a biopsy does not define how a lesion will behave.
The argument that identification of invasive cancer or the histologic subtype is critical is flawed. Evidence of a prognostic impact comes from studies that have included both GGNs and completely solid cancers. Applying these data to predominantly ground glass tumors is inappropriate. Those studies that have examined the behavior of predominantly ground glass tumors have demonstrated no difference in outcomes for invasive cancer versus AIS or MIA ( Fig. 4 A). ,
Can Baseline Imaging Characteristics Be Used?
Completely solid cIA lung cancers clearly pose a threat and thus warrant prompt treatment, even for tumors less than 1 cm. Node involvement is found in ∼10%, 20%, and 30% of solid cI T1a, T1b, and T1c tumors, respectively. , Despite resection, 5-year overall survival (OS) is approximately 85%, 70%, and 60% for cI T1a, T1b, and T1c tumors, respectively. , ,
However, most GGNs do not pose an imminent threat. Multiple studies demonstrate that a ground glass component correlates with excellent outcomes after resection, much better size-for-size or by T subgroup when compared with solid tumors (either overall or consolidated size). , , , , Several studies , , , have reported such excellent outcomes for resected GGNs—with little impact of the consolidated component size or proportion thereof—that it is unclear where an inflection point lies at which survival diminishes.
Pure GGNs consistently have excellent survival without recurrence, regardless of overall lesion size (but only rare cases are >3 cm). , , In a prospective study of lobectomy (Japan Cooperative Oncology Group [JCOG] 0201), 5-year recurrence-free survival (RFS) did not vary significantly by overall lesion size if the CTR was less than or equal to 0.25 (97% for cI ≤ 2 cm, 94% for >2–3 cm). Others have noted that overall tumor size has no impact, even with inclusion of some with higher CTR ratios ( Fig. 4 B).
Several studies suggest that cIA1 mixed GGNs remain highly curable even with consolidated components of 6 to 10 mm. For cI T1a mixed GGNs a 5-year OS of 97.5% was reported (n = 123, 45% invasive carcinoma, 39% lobectomy, 40% segmentectomy). This was confirmed in another study of mixed cIA1 GGNs (n = 102, mean consolidated size 8.3 mm on lung windows [2 mm slices], 71% invasive carcinoma, 55% lobectomy). No node involvement was noted for mixed cI T1a GGN tumors. , Recurrence has been noted in 1.6% of cIA1 mixed GGNs (on lung windows, 2 mm slices) and a 5-year RFS of 94% to 95%. ,
The data are less consistent for consolidation size of greater than 10 mm (lung windows) or greater than 50% of the total lesion size. Several studies have reported excellent 5- to 10-year OS for resected mixed GGNs, with no clear impact of the T subgroup (T1a, T1b, T1c) by consolidated component size ( Fig. 4 C). , Multivariate analyses of mixed GGNs have shown no prognostic significance of the consolidated component size in some studies. On the other hand, another study found a survival decrease by consolidation size in mixed GGNs (5-year OS 99%, 89%, and 89% and RFS 95%, 85%, and 72% for cIA1, cIA2, and cIA3, respectively). In addition, the incidence of node involvement for mixed GGN tumors was 3% and 14% for T1b and T1c tumors, respectively, in one study (by consolidation size on lung windows).
Several studies have reported excellent 5- to 10-year OS for resected mixed GGNs with no clear impact of the consolidated proportion (CTR ≤0.5 vs >0.5 ( Fig. 4 D). , , A small trend toward slightly inferior RFS was noted for CTR greater than 0.5 and larger consolidation size (20–30 mm). , Multivariate analyses of mixed GGNs have shown no prognostic significance of the CTR ratio in some studies. On the other hand, the incidence of node involvement for mixed GGN tumors is 0 versus ∼5% for CTR less than or equal to 0.5 versus greater than 0.5. , Others have noted decreasing survival for cIA mixed GGNs according to the CTR (5-year RFS of 99% and 91% for CTR ≤0.5 vs > 0.5). Furthermore, a study of large, mostly consolidated cN0 GGNs (average consolidated size 23 mm, CTR >0.75) found that although node involvement occurred in only 3% and outcomes were much better than completely solid similar-sized lung cancers, the 5-year OS was 87%. Taken together, it seems that a consolidated portion of greater than or equal to 10 mm or a CTR of greater than 0.5 in a mixed GGN warrants consideration of intervention.
Less data are available regarding the impact of a solid portion on mediastinal windows. Specific survival after resection by solid portion size has not been reported. One study noted little impact of the TDR on RFS after resection (mostly lobectomy) in mixed GGNs with an overall size less than or equal to 1 cm (specifically TDR of 0 to <0.4 vs 0.4–0.8, Fig. 5 ). This implies a 5- to 10-year RFS of ∼90% for small (≤1 cm overall) resected GGNs with a solid portion of ∼2 to 6 mm (on mediastinal windows). Larger mixed GGNs exhibited clearly worse survival, although there was still no clear difference between TDR categories. Taken together, a solid component on mediastinal windows that is less than ∼5 mm appears to result in excellent survival, but this may not be true with a larger solid component.
Another approach is using the 5 mm threshold that has (arbitrarily) been chosen as the dimension of invasiveness that differentiates MIA from invasive cancer as a surrogate endpoint. A detailed study (5 institutions, 15 radiologists, 378 patients, multiple blinded assessments) found that a size of 8 mm on lung windows and 6 mm on mediastinal windows correlated best with 5 mm on histologic examination (with ∼80% sensitivity and ∼80% specificity for each). These dimensions are roughly in alignment with the thresholds discussed in the preceding paragraphs. GGNs below these thresholds seem to pose little risk.
Can Assessment of How a Ground Glass Nodule Behaves Be Used?
The observation that most GGNs do not progress and the good outcomes after resection of GGNs suggest a surveillance strategy may be best to identify those GGNs that progress. This section addresses questions raised by such an approach.
An initial period of observation is usually indicated, because about 30% of GGNs resolve, typically within a few months. , However, the evidence considered in this chapter pertains to persistent GGNs (ie, that did not resolve).
In considering a surveillance approach, it is helpful to consider that the clinical and pathologic T stage category, respectively, is defined by the size of the solid (by imaging, CT parameters not specified) or invasive component, based on multiple studies demonstrating that this determines prognosis). This means that pure GGNs are classified as T0 and many mixed GGNs as Tis or Tmi. However, further analysis suggests that classification by solid/invasive tumor size does not bring GGNs and solid lung cancers prognostically into alignment—the presence of a ground glass component consistently results in better survival for each stage subgroup. , , , This effect is noted even with only a minor ground glass component (ie, CTR >0.5). , Furthermore, good OS and RFS is reported whenever a ground glass component is present, with the consolidated component size having relatively little impact.
Is a surveillance approach safe?
The safety of observation to assess for progression is a critical question if we adopt a surveillance strategy. The evidence presented earlier that most of the GGNs do not change, even over many years, and that change is gradual in those that do supports this approach. Most studies reporting rates of growth report only indolent growth, , , but some report rare cases (∼2%) with growth of ∼5 mm/y—although it seems this was not necessarily sustained and the subsequent outcomes of these patients is unknown. , It seems that a change from a pure GGN to a mixed GGN is a predictor of more rapid growth. This is consistent with a recent study that suggests that the VDT of the solid component in mixed GGNs is similar to that of traditional NSCLC (∼6 months).
Two prospective studies provide strong evidence that observation is safe ( Fig. 6 ). , These studies involved predominantly pure GGNs, with few having a CTR greater than 0.5 or a solid portion on mediastinal windows. Of the small subset of patients in these studies that eventually underwent resection, all were stage pIA, with less than 2% being T2aN0M0 due to visceral pleural invasion (which has unclear prognostic impact in small lesions). The average size of the solid component in one prospective study at the time of resection was 3.7 mm (mediastinal windows), and the average total size was 16 mm ; this was not reported in the other study. Subsequent follow-up of the resected patients found that all were cured in one study and 98.4% in the other. (Two patients [1.6%] in the latter study developed recurrence and died about 2.5 and 4 years postresection. Both exhibited increasing CTR over ∼3 years, beginning from a total size of 5 mm and a CTR of 0.75 in one patient and a 27 mm pure GGN in the other. Imaging characteristics of the tumors at the time of resection was not provided, leaving it unclear if earlier intervention could have altered the outcome.) Thus, prospective studies suggest that surveillance, with intervention when there is a change, does not result in stage progression or affect curability, with a rate of exceptions that is similar to the rate of major morbidity/mortality from anatomic pulmonary resection.