The profile of pleural effusions in oncological patients referred to the cardiothoracic surgery department in a tertiary hospital in Pretoria, South Africa: A retrospective study

Abstract

Background

The presence of a pleural effusion in any oncological patient (patient with known underlying current or previous malignancy) tends to be classified as advanced disease with few curative measures.

Objectives

To establish the aetiology of pleural effusions in oncological patients in our subpopulation. The primary objective was to determine the percentage of proven malignant pleural effusions (MPEs). Secondary objectives were to determine the causes of proven MPEs and proven benign (non-malignant) pleural effusions (non-MPEs), and to determine whether proven non-MPEs change the initial cancer staging of patients.

Methods

This was a retrospective record review of oncological patients admitted with suspected MPE to the Steve Biko Academic Hospital Cardiothoracic Surgery Unit during the period January 2018 - June 2020.

Results

Ninety-one patients met the inclusion criteria. The median (interquartile range) age was 56 (41 - 66) years, and there were 54 females (59.3%) and 37 males (40.7%). Pleural biopsy results confirmed MPE in 80.2% (95% confidence interval 70.6 - 87.8) of cases (n=73/91) and non-MPE in 19.8% (n=18/91). The aetiologies of MPEs in descending order were breast (42.5%), lung (24.7%), genitourinary tract (15.1%), lymphoma (9.6%), other (6.8%) and gastrointestinal tract (1.4%) carcinomas. The aetiologies of the non-MPEs were inflammatory (88.9%) and infective (11.1%) in nature. Among the proven non-MPEs, the results led to a possible cancer downstaging in 6/18 patients (33.3%).

Conclusion

Among oncological patients with pleural effusions referred to our unit, a statistically significant proportion (80.2%) had MPE by pleural tissue confirmation, with breast carcinoma being the most common cause. Of those with proven non-MPE, only a third had a possible cancer downstaging.

Study synopsis

What the study adds. This study adds to the growing local evidence-based research by various authors that can be compared with other local and international literature. It also adds to the information needed when establishing or revising protocols in oncological patients with pleural effusions.

Implications of the findings. In our population subgroup, the pleural effusions in oncological patients were mostly malignant in nature. The most common cause was breast carcinoma. This is not just a reflection of local referrals, as our unit also receives referrals from other provinces such as Mpumalanga, and sometimes Limpopo.

Background

Pleural effusion refers to the presence of fluid in the pleural cavity, and this accumulation can be either physiological or pathological. Malignancy is one of the causes of pathological pleural fluid accumulation, and it accounts for 20 - 30% of cases of exudative pleural effusions in adults.^[1] Malignant pleural effusions (MPEs) usually present in the disseminated and advanced stages of malignant disease.^[2] The presence of MPE implies that cancer has infiltrated the pleural tissue, with positive cytology and/or pleural biopsy.^[3] The mechanisms of this pleural infiltration vary. Patients with MPE have a poor prognosis, and the median survival is 3 - 12 months in most studies.^[4,5]

Confirmed MPE affects >150 000 patients in the USA and >40 000 in the UK each year.^[6] Lung and breast cancer account for 50 - 65% of cases; lymphoma, gastrointestinal tract (GIT) and genitourinary tract (GUT) tumours for 25%; and unknown primary tumours for 7 - 15%.^[7] In the South African (SA) context, Koegelenberg et al.^[8] found the causes of MPEs to be lung cancer, breast cancer, carcinoma of unknown origin and mesothelioma at a tertiary hospital in Western Cape Province.

HIV infection is prevalent in SA, and the immunosuppression also increases the risk of malignancy. However, other opportunistic infections such as tuberculosis can also occur, and they can cause pleural effusions. It is important to confirm the cause of pleural effusions in oncological patients, because non-malignant pleural effusions (non-MPEs) could indicate a more curable cause in some cases. The causes of pleural effusions in the present study were determined using pleural tissue biopsy. This method of diagnosis is in accordance with the 2018 European Respiratory Society/European Association for Cardio-Thoracic Surgery (ERS/EACTS) statement on the management of MPEs, which showed that literature supports the use of tissue biopsy as the gold standard for diagnosis and treatment planning.^[9]

Methods

Study design and unit setting

This was a retrospective cross-sectional study relying on the records of patients admitted to the Steve Biko Academic Hospital (SBAH) Cardiothoracic Surgery Unit from January 2018 to June 2020. SBAH is a tertiary healthcare institution that offers specialised services to patients referred from clinics and primary and secondary hospitals. It is the main teaching hospital attached to the University of Pretoria, and is situated in Capital Park, Tshwane, in Gauteng Province.

Study sample and population

The population consisted of oncological patients (patients with known underlying or previous malignancy) with pleural effusions referred to the Department of Cardiothoracic Surgery at SBAH for possible palliative surgical interventions such as pleurodesis and/or indwelling pleural catheter (IPC) insertion. The final sample size was determined by inclusion and exclusion criteria. Importantly, patients had to have had a pleural biopsy done in theatre under vision for confirmation of aetiology. Inclusion criteria were records of oncological patients with pleural effusions; all pleural effusion sites; records of patients taken to theatre for pleural biopsy and surgical intervention concurrently, or in the same hospital admission period; and patients of any age, gender and race. Patient records were excluded if there were no records of known underlying malignancy; if the patient had pleural effusions but no evidence of known underlying malignancy; or if the patient could not be taken to theatre owing to poor performance status, patient refusal, etc.

Procedures and data collection

The primary investigator (PML) collected data from the cardiothoracic surgery ward admission books, theatre records books, the hospital records department for patient files, and the National Health Laboratory Service (NHLS) website for histology results. It is our departmental policy to do pleural biopsies in the same setting when patients are taken to theatre for either pleurodesis or IPC placement. The pleural biopsy samples were analysed by the University of Pretoria Anatomical Pathology/NHLS department according to their processing guidelines.

Data management and statistical analysis

The data were collected with the use of data collection sheets, and the raw data were then recorded on an Excel worksheet, version 2016 (Microsoft, USA). The data were analysed by the biostatistician (PJB) using Stata version 18.0 BE statistical software (StataCorp, USA). Demographic and clinical parameters were summarised by reporting, but not limited to, the mean, standard deviation, median, range and 95% confidence interval (CI) for continuous data, and frequency, percentage and 95% CI for discrete data. The proportion of malignant biopsies was then expressed as a percentage along with a 95% CI. Following the results of the biopsies, the proportion of patients whose stage changed as a result was then determined. The study assessed patient records from January 2018 to June 2020, a period of 30 months. Theatre lists are restricted during December and January. The expected number of pleural effusions for which pleural biopsies were done would be between 75 and 250, i.e. 3 - 10 per month for 25 months (excluding December and January). The proportion of malignancies was expected to be very high, 80 - 90%. For an expected malignancy rate of 80%, a sample size of at least 75 pleural effusions would estimate the required proportion with 95% confidence to an accuracy within 9.1%. Furthermore, for an expected malignancy rate of 90%, a sample of 75 pleural effusions would estimate the required proportion with 95% confidence to an accuracy within 6.8%.

Ethical considerations

The study was approved by the University of Pretoria Research Ethics Committee (ref. no. 838/2020) and registered on the National Health Research Database. Permission was granted by the SBAH management and the University of Pretoria Department of Anatomical Pathology.

Results

Sample selection

A total of 244 patient records over a period of 30 months (January 2018 - June 2020) were audited. A total of 153 patients were excluded, leaving 91 patients eligible for inclusion (Fig. 1). This final sample size still fell within the estimated range of 75 - 250 patients.

Patients excluded were those who had a history of known primary malignancy and pleural effusion but could not undergo pleural biopsy (n=41), and patients with pleural effusion who underwent pleural biopsy but had no history of known underlying malignancy (n=112). Those with a known underlying primary malignancy who could not undergo pleural biopsy (n=41) did receive pleural fluid drainage in the ward together with talc slurry pleurodesis or tunnelled IPC insertion where indicated. Some of these patients were admitted in an already terminal state, and received end-of-life supportive care.

Among those who underwent pleural biopsy but had no known underlying primary malignancy (n=112), the following were found: incidental carcinoma (n=63, 56.3%); pleural tuberculosis (n=24, 21.4%); and other pleural biopsy findings (n=25, 22.3%) that were infective (empyema) and inflammatory (pleuritis).

Fig. 1.

Participant selection. (TB = tuberculosis.)

Sample characteristics

The characteristics of the study cohort (n=91) are set out in Table 1. The median (interquartile range) age was 56 (41 - 66) years. There were more females (n=54, 59.3%) than males (n=37, 40.7%) in the cohort. Based on pleural biopsy findings, 73 patients (80.2%) had malignant and 18 (19.8%) non-malignant disease. The MPEs were subdivided into six groups. With regard to surgical palliative procedures, most of the patients underwent IPC insertion (n=76, 83.5%), while chemical pleurodesis by talc poudrage was done in 13 patients (14.3%) and rapid pleurodesis (combination of IPC and chemical pleurodesis) in two. The sample size of 91 patients represents all eligible patients over the study period who met the inclusion criteria. While no formal power calculation was performed, this sample size provides adequate precision for the primary outcome, with a 95% CI width of ~17% for the MPE rate (Table 2).

Table 1. Study cohort characteristics (N=91).

Characteristic	n(%)*
Age (years), median (IQR)	56 (41 - 66)
Age groups (years)
10 - 19	1 (1.1)
20 - 29	2 (2.2)
30 - 39	17 (18.7)
40 - 49	18 (19.8)
50 - 59	20 (22.0)
60 - 69	21 (23.1)
70 - 79	10 (11.0)
≥80	2 (2.2)
Sex
Male	37 (40.7)
Female	54 (59.3)
Malignancy	73 (80.2)
Breast	31 (42.5)
Lung	18 (24.7)
Lymphoma	7 (9.6)
GUT	11 (15.1)
GIT	1 (1.4)
Other	5 (6.8)
Non-malignant	18 (19.8)
Surgical palliative procedures
Pleural biopsy + IPC	76 (83.5)
Pleural + chemical pleurodesis	13 (14.3)
Pleural biopsy and rapid pleurodesis	2 (2.2)

IQR = interquartile range; GUT = genitourinary tract; GIT = gastrointestinal tract; IPC = indwelling pleural catheter.

* Except where otherwise indicated.

Table 2. Percentage of malignant and non-malignant pleural effusions (N=91).

Malignant	Frequency	Percentage	Cumulative
No	18	19.8	19.78
Yes	73	80.2	100
Variable	Observations	Proportion	Standard error	Confidence interval
Malignant	91	0.8021978	0.0417576	0.705523 - 0.8783501

Descriptive statistics for study cohort

Among the 91 patients who met the inclusion criteria, pleural biopsy confirmed that 80.2% (n=73/91) had malignant pleural effusions (95% CI 70.6 - 87.8) (Table 2).

In proven MPEs, breast carcinoma was the most common cause (42.5%), followed by lung (24.7%) and GUT (15.1%) carcinomas, lymphoma (9.6%), and other (6.8%) and GIT (1.4%) carcinomas, in descending order (Fig. 2). The ‘other’ group was made up of five patients, who had Kaposi’s sarcoma (n=1), Hürthle cell thyroid carcinoma (n=1), chronic lymphocytic leukaemia (n=2) and malignant mesothelioma (n=1). The causes of non-MPE were inflammatory (88.9%) and infective (11.1%) in nature (Fig. 3). In the group with proven non-MPE (Table 1), the initial cancer staging changed for six patients (33.3%) after the pleural biopsy results (Table 3).

Fig. 2.

Aetiologies of proven malignant pleural effusions. (GUT = genitourinary tract; GIT = gastrointestinal tract.)

Fig. 3.

Aetiologies of non-malignant pleural effusions.

Table 3. Stage changes in non-malignant pleural effusions (N=18).

Primary malignancy	n(%)	Number of original stage changes after the pleural biopsy
Breast	5 (27.8)	0
Lung	3 (16.7)	0
Lymphoma	5 (27.8)	4
GUT	2 (11.1)	0
GIT	0	0
Other	3 (16.7)	2

GUT = genitourinary tract; GIT = gastrointestinal tract.

Discussion

A total of 91 patients were included in the analysis, with a median age of 56 years and a predominance of females (59.3% v. 40.7%), as seen in Table 1. Taghizadeh et al.^[10] in the USA reported a median age of 68 years, but also a female predominance (55.8%). The majority of pleural effusions in the present study were proven to be malignant (80.2%), and the CI suggests a reasonably precise estimate of this population (Table 2). The diagnostic modality was pleural biopsy under vision in theatre via thoracoscopy, or open pleural biopsy in some cases. Thoracoscopy has been reported to have a superior diagnostic yield for malignancy, with a sensitivity of 96.7%.^[11]

Fig. 2 shows that the most common cause of MPE in this population was breast carcinoma (42.5%). This figure is similar to that reported in an Egyptian study by Allama et al.,^[12] in which breast carcinoma was the most common cause (43.6%) in the malignant pleural effusion group. In a retrospective study by Nwafor et al.^[13] in Nigeria, metastatic breast cancer was the most common cause of MPE, followed by advanced lung cancer. A potential reason why breast cancer could have been the most common cause of MPE in our cohort is that breast cancer tends to have relatively good screening protocols locally, and patients with lung carcinoma often present with advanced disease and in poor physiological states. Koegelenberg et al.^[8] found lung carcinoma to be the most common cause at their institution in Western Cape Province, SA, and this correlates with most of the published international literature.^[14–18] With regard to MPE gender cohorts in the present study, breast carcinoma was most common in females (42.5%) and lung carcinoma most common in males (15.1%). International literature also shows MPE caused by lung cancer to be more common in men, and MPE caused by breast cancer to be more common in women.^[19,20]

In the 18 patients with proven non-MPE, the causes were infective (11.1%) and inflammatory (88.9%), as shown in Fig. 3. The infective group was attributed to tuberculosis, while the inflammatory group was attributed to chronic/fibrosing/nonspecific pleuritis.

Table 3 shows that the initial cancer staging changed in only six patients (33.3%) after their pleural biopsies. These patients had histories of completely treated lymphoma (n=4), completely excised bilateral lacrimal gland tumour (n=1), and completely excised malignant melanoma of the chest wall (n=1). The pleural biopsy results in these patients were tuberculosis (n=2), which is a curable condition, and chronic pleuritis (n=4). The presence of malignant tumour cells in the pleural space upstages a patient’s primary cancer to metastatic, which is associated with high morbidity and low survivability, with a median survival of 4 - 9 months.^[21] In cases where downstaging is achieved, it does positively affect the patient’s biopsychosocial spectrum.

Incidental carcinoma was found in 56.3% (n=63) of the 112 patients who underwent pleural biopsy but were not eligible for inclusion in the final cohort. It has been noted in the literature that up to 15% of patients have MPE diagnosed incidentally, and pleural effusions can be their only presenting sign of cancer.^[22] Another notable finding was pleural tuberculosis in 21.4% (n=24) of the patients (Fig. 1). This is not an unusual finding, as tuberculosis is endemic in SA, and the local rate of pleural tuberculosis has been reported to be as high as 30%.^[23]

The surgical palliative procedure that was performed most frequently was IPC insertion (83.5%) due to patients having non-expandable lungs or inadequate lung expansion for pleurodesis. IPC can achieve symptom control comparable to chemical pleurodesis, improved quality of life, shorter hospital stays and fewer further pleural interventions, and because of these benefits it has been noted in some literature as the preferred initial first step in MPE with either an expanded or a non-expanded lung.^[3]

Study limitations

This study was retrospective in nature. Obtaining high-quality records was a limitation in some cases, and various sources were consulted to gather sufficient data to ensure that the study findings were a true reflection of reality and of an acceptable standard. The final sample size (n=91) was small but still within the biostatistical estimation, although a larger sample size over a prolonged period would perhaps have provided additional information that could have supplemented the study to give it more power. The pleural biopsies were performed by different surgical registrars in our unit, and not by a single specified person. This was a single-centre study with exclusive selection bias. The results were based mainly on pleural histopathological findings, and other parameters such as pleural fluid cytology and biomarkers were not included.

Conclusion

The majority of the pleural effusions in oncological patients referred to our unit are malignant in nature (80.2%), with breast carcinoma being the most common cause. A minority of patients (19.8%) had proven non-MPE that was attributed to pleuritis and tuberculosis, and of these patients only one-third (33.3%) achieved possible cancer downstaging.

Based on the above, it is justifiable to perform palliative pleural procedures in oncological patients with suspected MPE, even in cases where pleural biopsy tissue confirmation is not available or attainable.

As mentioned by Shaw et al.,^[24] who are based in the Western Cape, SA, the incidence of MPE in low- and middle-income countries is not known. Our recommendation is to perform more studies on malignant pleural effusions in SA, not only to contribute to the global literature, but also because MPE continues to be a burden on society and all efforts should be made to improve patients’ quality of life.

As stated by Thomas et al.,^[15] ‘We must however remember that drainage no matter how efficient, is not the ultimate therapy for MPE. The holy grail in MPE management is to control and stop the effusion from forming in the first place. To achieve that, we will need to expand our very limited knowledge on the pathobiology of MPE formation.’

In our various efforts to better understand and manage MPE, we should always remember the biopsychosocial spectra and the importance of adhering to the four core principles of medical ethics, one of which is patient autonomy. This respect for patient choices was highlighted in an expert review by Fitzgerald et al.,^[25] who pointed out that patient preference must be considered in deciding on what treatment is chosen, given that the primary focus of MPE care is to improve quality of life.