Lung cancer tumor markers in serous effusions and other body fluids

1Introduction

Lung cancer (LC) is the third most common cancer in the United States. According to the American Cancer Society, 236,740 new cases are expected in the general population in 2022, with an estimated mortality rate of 130,180 cases [1]. In 2020, there were more than 2.2 million new cases of LC worldwide with a forecast of 1.8 million deaths.

In Europe the data are similar to those of the US. In 2020, 2,339,617 cases of lung cancer were recorded, representing 13.5% of tumors among men (315,054 cases) and 7.9% among women (162,480 cases) [2]. Lung cancer is the leading cause of death due to cancer in men and the third highest in women [3]. It rose from fourth to third place among women in 2019, probably due to the increase in tobacco consumption in women since 1970 [4].

The severity of the prognosis of LC means that the clinical practices for the treatment of the disease need to be continuously updated. Today, due to advances in the understanding of specific molecular pathways, it is possible (and also essential) to differentiate between adenocarcinoma, squamous cell carcinoma, and other histological types of LC. Thanks to this new knowledge, targeted therapies can be designed which incorporate drugs that are effective against particular types of tumor but may be toxic in other contexts (for example, tyrosine kinase inhibitors).

The 2015 World Health Organization (WHO) classification reflects the distinction between types of LC [5].

2Lung cancer and effusions

An effusion is a pathological accumulation of fluid in the space between the visceral and parietal serosa (the pleural or pericardial space). In cancer patients there are two main types:

Effusions in LC may have different etiologies:

3Diagnosis of neoplastic effusions

It is important to correctly diagnose pleural fluids as malignant or non-malignant according to the TNM classification of the International Association for the Study of Lung Cancer (IASLC) (8th edition, 2015) [13]. The presence of malignant pleural or pericardial effusion is associated with low patient survival. Patients with stage I/III lung cancer and benign serous effusions (pleural and/or pericardial) are candidates for tumor resection, and have a better prognosis than patients with malignant effusion even with the same TNM; the latter are considered metastatic (M1a), and are non-operable [14].

Therefore, an accurate differential diagnosis in these patients is essential in order to establish prognosis and select treatment. The simplest definitive method for diagnosing a neoplastic effusion is pleural fluid cytology. Between 30% and 50% of cytologies are negative, and percutaneous pleural biopsy increases the sensitivity for detecting malignant pleural effusion by 10%. Thoracoscopy has additional diagnostic value in undiagnosed cases of exudative pleural effusion with high clinical suspicion of malignancy, but it is a highly invasive procedure. Therefore, alternative methods are necessary to be able to select patients for these tests.

4Tumor markers and effusions

5Inflammatory processes and tumor markers: False positives

Some benign processes with high levels of inflammation or necrosis of the serosa and surrounding tissues, such as empyema and tuberculous or complicated parapneumonic effusions, can present high TM concentrations in serous fluid. The levels may reach up to 10 times the concentrations of benign effusions without inflammation.

Excluding purulent effusions, Porcel et al. (2017) obtained a slight improvement in sensitivity [41% (45 ng/mL) vs 31% (50 ng/mL) for CEA and 40% (77 IU/mL) vs 37% (75 IU/mL) for CA15-3] with cut-off points similar to or slightly lower than when these effusions are included [21].

These inflammatory effusions are easily identifiable using routine laboratory biomarkers such as adenosine deaminase (for tuberculous effusions or empyema), differential leukocyte count (for empyema or complicated parapneumonic effusions) or C-reactive protein as a marker of inflammation (for tuberculous effusions, empyema, parapneumonic effusions, and severe inflammation) [22, 23], which we can term markers of benignity. These tests allow us to identify false positives and to apply higher discriminant values in patients with positive biomarkers of benignity, thus improving the specificity.

Another cause of high values of TMs in serous fluids is the passage from serum to fluid through the membrane, either due to the presence of tumors that do not invade the mesothelium or other pathological processes such as kidney failure or liver disease that raise serum TM concentrations [22]. These possible false positives can be detected by the fluid/serum (F/S) ratio [24]. Patients with neoplastic invasion of the mesothelium will present higher concentrations in fluid than in serum, while in patients with elevated concentrations in serum the F/S ratio for markers that are usually produced by the mesothelium will be below 1 : 1.

Applying these criteria, our group has evaluated two strategies: one based on a high cut-off value (CEA = 60 μg/L, CA15-3 = 80 KU/L; CA19-9 = 201KU/L and CA72-4 21 KU/L) and the other on a low cut-off point and F/S ratio (CEA = 5 μg/mL; CA15-3 = 30 KU/L, CA19-9 = 37 KU/L and CA72-4 = 6.7 KU/L and F/S>1.2), seeking maximum specificity [16]. The sensitivity and specificity of more than 400 pleural effusion samples were assessed, obtaining a sensitivity of 64% for the high cut-off point and 77% for the low cut-off point and F/S ratio, with specificities of 100% and 98% respectively.

When classifying them according to the biomarkers of benignity (ADA, CRP and % of polynuclear cells) in the group with negative values, the low cut-off point and F/S ratio strategy showed a sensitivity of 80% and a specificity of 100%, and for the high cut-off point the sensitivity was 67% at maximum specificity. In both strategies, the positive predictive value (PPV) was above 99%. For the group with positive biomarkers of benignity, the sensitivity fell in both strategies; maximum specificity was maintained in the high cut-off point strategy, but fell below 95% for the low cut-off point and F/S ratio, with a PPV of 68.8%.

To facilitate the interpretation of TMs in serous fluids, Figure 1 describes an algorithm based on the experience with pleural effusions. The cut-off points established for specificities of 100% in different studies by the same group may be different, so the PPVs may not reach 100%. For this reason, the PPVs are shown as greater than 99%.

Fig. 1

Algorithm for interpretation of TMs in serous fluids [40].

6Pericardial effusion

Very few studies have assessed tumor markers in pericardial fluid [22]. Szturmowicz et al. found higher concentrations of CEA and CYFRA 21.1 in malignant pericardial effusions than in non-malignant ones, obtaining values of 80 ng/mL and 0.5 ng/mL respectively for CEA and 260 ng/mL and 22.4 ng/mL for CYFRA 21.1. In view of their results, they proposed optimal cut-off values for these TMs of 5 ng/mL for CEA and 100 ng/mL for CYFRA 21.1 [23].

In previous work the same authors studied non-specific enolase (NSE), obtaining values of 41.8 ng/mL. They argued against its evaluation in blood samples [25].

Karatolios et al. also found higher values in malignant pericardial effusions with specificities of 100% for the following TMs:

[cut off (sensitivity) CEA, 8.5 ng/mL (48%), CA 19.9, 856(15%), CA 72.4, 19.5 UI/mL (45%); SCC, 216 ng/mL (4%), NSE, 348.5 ng/mL (7%)] [26].

The European Society of Cardiology guidelines [27] recommend cut-off points of 5 ng/mL for CEA and 100 ng/mL for CYFRA 21.1.

7Cerebrospinal fluid

Approximately 20% of patients with lung cancer may have metastatic involvement of the central nervous system, although a few of them, around 10% [28], have leptomeningeal involvement, that is neoplastic meningitis (NM). Glass et al. reported that patients with brain metastases without leptomeningeal involvement did not present tumor cells in CSF [29]. The most frequent histological types in NM are SCLC and adenocarcinoma. Other tumors that can produce meningeal metastases are breast tumors, melanomas, lymphomas and leukemias [30, 31]. The diagnosis is made by CSF cytology, but the sensitivity is only 40–60% [32, 33].

Some TMs in cerebrospinal fluid have been studied as aids to diagnosis. As in the case of serous effusions, the discriminant values vary notably from study to study. In general, the CSF concentrations of non-neoplastic meningitis for most TMs (CEA, CA15-3, CA19-9, CA125 and CA72-4) are undetectable, while CSF concentrations of other markers such as NSE and S100 are low and indicate damage to the CNS [34, 35], and so false positives are possible. Table 4 displays the TMs and the discriminant values proposed by various authors.

In addition, increases in CSF TM concentrations due to leakage and/or dysfunction of the blood-brain barrier, or contamination via the blood due to traumatic lumbar puncture, may not have been correctly assessed [36, 37]. Therefore the intrathecal synthesis of the TMs should be measured, in the same way as other macromolecules such as specific IgG in infectious diseases or light chains in multiple sclerosis [31, 36]. Zhang et al. used cut-off points of 1.145 ng/mL for CYFRA 21-1 and 0.279 ng/mL for CEA, finding higher sensitivities and specificities than using the F/S ratio. For their part, Corsini et al. determined CEA, CA125, CA15-3 and CA19-9 in CSF and serum and also the intrathecal synthesis of TMs [38]. The sensitivity of intrathecal synthesis was 94.4% for CEA, 50% for CA125, 83.3% for CA15 -3 and 66.7% for CA19-9. Patients with NM presented intrathecal synthesis of at least one of these markers, but none of the patients with non-neoplastic neurological diseases showed intrathecal synthesis of any TMs; therefore, this strategy seems more useful to be able to identify the origin of the malignant meningitis.

8Conclusions

Lung cancer patients may present serous effusions due to different etiologies, either cancer- or non-cancer-related.

In lung cancer, invasion of the pleural and pericardial serosa by tumor cells is a frequent event. This invasion is more common in adenocarcinomas and SCLC than in other histological types.

Determining whether or not the effusion is malignant is important, given the implications for therapy. An error can lead to overtreatment (surgery when the effusion is malignant and is considered benign) or undertreatment (no surgery when it is benign and considered malignant), which can lead to a worse prognosis.

In the differential diagnosis of serous effusions, the cut-off points described for each of the TMs vary widely. The diagnostic performance of TMs is good when high cut-off points are used to obtain specificities greater than 99%, and combinations of TMs are used to increase sensitivity.

The diagnostic yield can be improved by identifying possible false positives and using specific cut-off points for each group, obtaining higher sensitivity for specificities above 99%.

Leptomeningeal metastases can occur during the progression of lung cancer, with ADC and SCLC being the most frequently involved tumors.

In non-neoplastic neurological processes, the concentrations of some TMs are usually undetectable. For correct interpretation, it is important to confirm the integrity of the blood-brain barrier and bear in mind its possible contamination due to traumatic lumbar puncture. The determination of the intrathecal synthesis of the markers may be a good alternative.