Bedside ocular ultrasonography for diagnosing increased intracranial pressure in patients with leptomeningeal metastases from non‐small‐cell lung cancer

Abstract Objectives To explore the diagnostic accuracy of ultrasound measurement of optic nerve sheath diameter (ONSD) and optic disc height (ODH) in detecting intracranial hypertension in non‐small‐cell lung cancer (NSCLC) patients with leptomeningeal metastases (LM). Methods Seventy‐two patients with NSCLC‐LM and 65 patients with NSCLC were enrolled. The ONSD, ODH, eyeball transverse diameter (ETD), and eyeball vertical diameter (EVD) were measured by ultrasound. Subsequently, lumbar puncture was performed in NSCLC‐LM patients to measure cerebrospinal fluid pressure (CSFP), and intrathecal chemotherapy was regularly implemented. Pearson's correlation analysis was used to analyze the relationship between CSFP and ultrasound findings. The diagnostic accuracy of ONSD, ODH, and combined ONSD and ODH was evaluated by receiver operating characteristic (ROC) curve analysis and the corresponding area under the ROC curve (AUC). Results The ONSD, ODH, ONSD/ETD, and ONSD/EVD values were higher in the NSCLC‐LM group (all p < 0.05). The ONSD, ODH, ONSD/ETD, and ONSD/EVD values were all elevated in the abnormally elevated CSFP group (all p < 0.05). ONSD, ODH, ONSD/ETD, and ONSD/EVD were positively correlated with CSFP (r = 0.531, 0.383, 0.534, and 0.535, all p < 0.0001). The AUCs for ONSD, ODH, ONSD/ETD, and ONSD/EVD to detect CSFP >280 mmH2O were 0.787 (95% CI: 0.64–0.93, sensitivity 68.75%, specificity 91.07%), 0.885 (95% CI: 0.81–0.96, sensitivity 100%, specificity 69.64%), 0.765 (95% CI: 0.64–0.89, sensitivity 81.25%, specificity 64.29%), and 0.788 (95% CI: 0.64–0.93, sensitivity 56.25%, specificity 91.07%), respectively. When ONSD was combined with ODH, the AUC was 0.913 (95% CI: 0.83–0.99, sensitivity 87.85%, specificity 85.70%). Furthermore, intrathecal chemotherapy was associated with a downtrend in CSFP and ultrasound findings. Conclusion There are important advantages of using bedside ultrasonography for detecting elevated CSFP in NSCLC‐LM patients. Further research should be performed to evaluate the clinical significance of an enlarged ONSD and increased ODH in NSCLC‐LM.


| INTRODUCTION
In the advanced stage of non-small-cell lung cancer (NSCLC), 3%-4% of patients develop leptomeningeal metastasis (LM). 1,2 According to previous research, the median overall survival (OS) of NSCLC patients with LM is dismal, only 3 months. 3 Timely diagnosis and treatment can improve the prognosis of patients. Lumbar puncture is a vital means for the diagnosis of meningeal metastasis, cerebrospinal fluid genomics, and intrathecal chemotherapy. However, NSCLC-LM usually leads to intracranial hypertension, and lumbar puncture carries a high risk of cerebral hernia in such patients. Monitoring and management of intracranial pressure (ICP) is integral before lumbar puncture, and invasive ICP monitoring is the gold standard. However, the placement of invasive ventricular devices could be challenging due to the risk of infection and bleeding, lack of surgical availability, or high cost. Therefore, a noninvasive, bedside and easy-to-use detection tool is urgently needed to detect elevated ICP.
The optic nerve sheath (ONS), which surrounds the optic nerve, is separated by the subarachnoid space and filled with cerebrospinal fluid. The ONS is a continuation of the dura. 4 When ICP is elevated, ONS is distensible in its retrobulbar segment. 5 Moreover, prior reports have demonstrated that optic disc edema can be observed in chronic intracranial hypertension, 6 which is due to swollen nerve fibers and the accumulation of extracellular fluid. US measurements of ONSD have been studied in different clinical settings and have shown a good correlation with invasive ICP values. These studies suggest that the ultrasonic measurement of ONSD is a reliable noninvasive method for assessing intracranial pressure. [7][8][9] Under dynamic conditions, the correlation between ONSD and ICP remains valid. 10 Of the studied ultrasound noninvasive ICP methods, ONSD, when compared with transcranial Doppler (TCD), showed higher accuracy in estimating ICP. 11 However, a recent meta-analysis showed that the optimal ONSD threshold varies greatly between studies. 4,9,12,13 Meanwhile, it is not clear whether ODH or ONSD combined with ODH can evaluate intracranial hypertension in diagnosing elevated intracranial hypertension. Previous studies have mainly focused on traumatic brain injury, cerebrovascular accidents, and central nervous system infection, but the application of ONSD ultrasound in lung cancer meningeal metastasis has not been reported.
Our main purpose was to recognize the accuracy of ultrasonographic measurements for the detection of cranial hypertension in patients with LM from NSCLC and to identify the optimal ONSD and ODH threshold.

| Ethics statement
We conducted a retrospective cohort study using data from the Department of Geriatric Oncology in Nanjing Drum tower Hospital in China. We signed informed consent with family members of all patients. This study was authorized by the Ethics Committee of Nanjing Drum tower Hospital. All procedures have proceeded in line with the ethical principles of the Declaration of Helsinki.

| Patients
Between March 2021 and December 2021, 137 patients (72 with LM and 65 without LM) with NSCLC in the Department of Geriatric Oncology in Nanjing Drum tower Hospital were enrolled. All patients were diagnosed by clinical, CSF and radiographic findings based on the EANO-ESMO criteria. 14 The inclusion criteria were as follows: (1) 18 years or older, (2) diagnosis of lung adenocarcinoma, and (3) patients with positive CSF cytology or with negative CSF cytology but with typical MRI and clinical symptoms requiring lumbar puncture and intrathecal chemotherapy. The exclusion criteria were as follows: (1) the use of medications that affect intracranial pressure in 1 week, (2) bilateral optic nerve involvement and intrathecal chemotherapy was associated with a downtrend in CSFP and ultrasound findings.

Conclusion:
There are important advantages of using bedside ultrasonography for detecting elevated CSFP in NSCLC-LM patients. Further research should be performed to evaluate the clinical significance of an enlarged ONSD and increased ODH in NSCLC-LM.

K E Y W O R D S
cerebrospinal fluid pressure, leptomeningeal metastases from non-small-cell lung cancer, ocular ultrasonography preexisting ocular disease other than ametropia, and (3) recent use of mannitol and other drugs that affect intracranial pressure. The subjects in our control group were mainly postoperative adjuvant chemotherapy patients (pathological stage IIA-IIIA), who had no neurological signs and symptoms and whose cerebrospinal metastases were ruled out by cranial and whole spine MRI.

| Clinical data collection
General information were recorded at admission: sex, age, BMI, oxyhemoglobin saturation, blood pressure, heart rate, and blood glucose. All patients were diagnosed by clinical symptoms, radiographic findings and cerebrospinal fluid cytology before enrollment. Two experienced ultrasound physicians performed ultrasound measurements at the bedside. Here, a 5-to 12-MHz linear array ultrasonic probe from Philips iU22 (Philips Healthcare, USA) was used. The patient was supine, facing forward, and looking forward with the eyelids closed. A small volume of ultrasound gel is applied to the probe. And the probe was placed horizontally above the transverse axis of the eyeball with no pressure ( Figure 1A). Slowly and vertically moved to the eyeball until the clearest and artifact-free optimal plane was captured. The hypoechoic stripe posterior to the eyeball is the optic nerve, with sheaths on both sides visible and well-defined. Images were chosen by two expert investigators together. The optic nerve sheath could be seen at the broadest part; within 1 mm from the eyeball, the optic nerve could be seen continuously for 6 mm without motion artifacts. Calipers were used to determine the point 3 mm posterior to the globe, and the ONSD of that point was measured in the vertical plane. The maximum diameter of the ETD (eyeball transverse diameter) and EVD (eyeball vertical diameter) on this plane were also measured ( Figure 1B). Both eyes were measured three times, and the three values were averaged as the ONSD, ETD, and EVD of each eye. ODH was measured between the retina and the optic disc dome ( Figure 1C). 15 If no elevation of the optic disc was observed, ODH was recorded as 0 mm. The mean ONSD, ETD, EVD, and ODH of the patient were obtained by averaging the ONSD, ETD, EVD, and ODH of the bilateral eyes. The optic nerve ultrasound was performed about half an hour before the lumbar puncture in all patients in our enrollment. The optic nerve ultrasound took approximately 5 min to complete. Subsequently, the clinician performed a lumbar puncture in NSCLC-LM patients, the CSFP was measured, and intrathecal chemotherapy was performed. The ultrasound operators and investigators were blinded to the patient's CSFP. We defined CSFP >280 mmH 2 O as abnormally elevated intracranial pressure. 16 The CSFP was recorded by another clinical doctor.

| Statistical methods
Use a t-test (for data with a normal distribution) or Mann-Whitney Utest (for data with a nonnormal distribution) for continuous variables when exploring the differences between NSCLC patients with and without LM and between LM patients with CSFP ≤280 mmH 2 O and CSFP >280 mmH 2 O. We implemented Pearson's correlation analysis to analyze the relationship between CSFP and optic nerve examinations. A receiver operating characteristic (ROC) curve was plotted to figure up the area under the ROC curve (AUC). All statistical analyses and graphics were performed with SPSS v26.0 (IBM) and GraphPad Prism v8.3.0 (GraphPad). A value of p < 0.05 (two-sided) was regarded significant.

| RESULTS
A total of 137 NSCLC patients were enrolled in our study, 72 of whom were identified with LM. The baseline  Table 1. The ONSD measurements of patients without LM and with LM were 0.45 (0.43-0.48) cm and 0.57 (0.54-0.61) cm, respectively. No optic disc edema was present in patients without LM, while 33/72 (45.83%) patients with LM were observed to have optic disc edema. The average ONSD measurement and the proportion of optic disc edema in patients with LM were significantly increased compared with patients without LM (p < 0.05).
We divided patients with LM into two groups based on where papilledema was observed (Table 3). We found that the ONSD value, ONSD/ETD and ONSD/EVD were significantly higher in the group with papilledema than in the group without papilledema (p < 0.001, p = 0.012, p = 0.003). We also divided patients with LM into two groups based on whether they had symptoms of intracranial hypertension, whether MRI showed lateral ventricular widening, whether brain metastasis was present, whether MRI showed meningeal enhancement, and T A B L E 1 Baseline variables and outcomes whether CSF had atypia cells. However, the ONSD value, ONSD/ETD and ONSD/EVD did not differ between the two groups. According to the applied Pearson's correlation analysis, ONSD, ODH, ONSD/ETD, and ONSD/EVD were positively correlated with CSFP (r = 0.531, p < 0.001; r = 0.383, p = 0.025; r = 0.534, p < 0.001; r = 0.535, p < 0.001). Scatter diagrams are presented in Figure 2.
The ROC curve analysis found that the AUCs of ONSD, ONSD/ETD, ONSD/EVD, ODH, and combined ONSD and ODH in diagnosing abnormally elevated CSFP of NSCLC patients with LM were 0.787, 0.765, 0.788, 0.885, and 0.913, respectively. When the ONSD value was 0.615 cm, the AUC was 0.787, and the sensitivity and specificity were 68.75% and 91.07%, respectively. The ROC curves of ONSD/ETD, ONSD/EVD, ODH, combined ONSD and ODH are shown in  Table 4 and Figure 3 for details.
Twenty-two patients with LM received six intrathecal chemotherapies. Intrathecal pemetrexed at a dose of 30 mg was administered every 21 days via lumbar puncture on days 1 and 8. 17 Half of them (50%; n = 11/22) showed a baseline ONSD>0.615 cm, and of the 22 patients, a baseline CSFP>280 mmH 2 O was present in 11 patients (11/22; 50%) on admission. We discovered that the initiation of intrathecal chemotherapy was associated with a downtrend in CSFP and ultrasound findings, including ONSD, ODH, ONSD/ETD, and ONSD/EVD (Figure 4). There was an apparent decline in the proportion of patients with widened ONSD and in the number of patients with abnormally elevated CSFP following each intrathecal chemotherapy ( Figure 5). From a baseline value of 50%, the proportion of patients whose ONSD was >0.615 cm decreased to 31.8% after the second intrathecal chemotherapy and to 13.6% after the fifth intrathecal chemotherapy. From a baseline value of 11, the number of patients whose CSFP >280 mmH 2 O decreased to eight after the second intrathecal chemotherapy and to zero after the fifth intrathecal chemotherapy.

| DISCUSSION
NSCLC-LM often leads to an elevation of intracranial pressure, which could severely disrupt nerve function and possibly even lead to a cerebral hernia, squeezing important structures and inducing coma or death in severe cases. Early recognition and management can improve the outcomes of LM patients. 18 However, invasive ICP monitoring is the "gold standard". 19 Rapid and bedside identification of NSCLC-LM patients requiring rapid control of ICP may drive effective management in an urgent or resource-constrained environment. 12 In previous studies, researchers have assessed only ONSD as a monitor for intracranial hypertension; now, we demonstrate additional optic nerve abnormalities with optic disc elevation. Our results indicate that ultrasonic ONSD and ODH are strongly related to CSFP. With the    decrease in CSF pressure, ONSD also showed a downward trend. The measurement of ONSD and ODH could be used to detect abnormally increased CSFP in NSCLC-LM patients. This is the first time we have used optic nerve ultrasound to predict CSFP associated with metastatic meningeal carcinoma.
There have been some studies on ultrasonic measurement of ONSD for the diagnosis of high ICP, 20,21 while there are relatively few studies on the measurement of ODH for the diagnosis of high ICP. Rajajee V et al. compared the accuracy of optic nerve ultrasound in assessing intracranial pressure with sharp fluctuations and stability. They found that the specificity and positive predictive value of ONSD measurement prominently reduced in the setting of acutely ICP fluctuation because of the delayed reversal of ONS dilation. 22 Hayreh SS et al. explored the pathogenesis of optic disc edema by performing basic, experimental, and clinical studies. They found that acutely elevated CSFP did not produce optic disc edema, while chronically elevated CSFP can lead to optic disc edema. 6 Therefore, ONSD and ODH are of great value in assessing CSFP in NSCLC patients with meningeal metastasis, which is subacute or chronic. In our study, ONSD and ODH were positively correlated with CSFP. ONSD values significantly increased in NSCLC-LM groups, especially in patients with CSF pressure greater than 280 mmH2O. More optic disc edema was available in the group with abnormally high CSFP. These discoveries confirmed the biological relationship between ONSD, ODH, and CSFP. However, in clinical practice, the presence of a biological correlation does not automatically translate to satisfactory diagnostic accuracy. As ONSD and ODH are the most potential noninvasive monitors to recognize patients requiring empirical treatment, high sensitivity is foremost. Meanwhile, a test with high specificity can be superior to routine clinical evaluation and is likely to outperform invasive testing. In our study, the optimal threshold of ONSD obtained only a sensitivity of 68.75% (along with a minimal specificity of 91.07%), which could not meet our predetermined minimum criteria for acceptable accuracy. With this threshold, nearly onethird of patients with truly intracranial hypertension would be suffering from error classification, which is unacceptable given the severe consequences of elevated intracranial pressure if left untreated. In addition, ODH achieved an AUC greater than 0.8 for the detection of concurrent CSF pressure greater than 280 mmH 2 O, but the optimal threshold of ODH achieved a specificity of 69.64% (in conjunction with minimal sensitivity 100%), which means that nearly one-third of patients diagnosed with elevated ICP using this ODH threshold will, in fact, not have intracranial hypertension and may be exposed to the risks of overtreatment. ONSD and ODH both performed badly in recognizing patients requiring an aggressive therapy to treat intracranial hypertension. However, when ONSD was combined with ODH, the optimal threshold achieved a specificity of 85.7%, in conjunction with a sensitivity of 87.5%. Therefore, we could use the combined ONSD and ODH to detect abnormally elevated CSF pressure in NSCLC-LM patients.
A recent study suggests that the extent to which ONS swell with increased intracranial pressure may vary from person to person. 23 Therefore, to increase the accuracy of the optic nerve ultrasound method, ONSD/ETD and ONSD/EVD were measured to evaluate their relationship with CSF pressure. Our study suggested that these variables may be utilized as a possible alternative, which is consistent with other studies. 16,21,24 However, we still need more investigations before using the ONSD/ETD ratio and other variables for clinical decision-making. Our study has several strengths. This is the first time we have used ultrasound to assess CSF pressure in NSCLC-LM patients. This is the only study to assess the ability of combined ONSD and ODH to identify patients requiring timely treatment of abnormally elevated ICP. Continuous changes in ultrasound findings and CSF pressure were observed before and after treatment. Most importantly, sonographers remain completely and rigorously blinded to the CSF pressure and clinical details of the patients. The standards for reporting diagnostic accuracy recognize blinding as a vital element of high-quality diagnostic studies. 25 There are still several limitations in our study. A larger sample size may have narrowed the CIs in the estimates of accuracy-there were only 16 (22.22%) CSF pressure > 280 mmH2O in the analysis. The ECOG score for all patients was between 0 and 2; therefore, the most severe cases were likely excluded. Moreover, we compared ultrasound values with lumbar CSF pressure instead of invasive ICP measurement, yet there may be differences in the CSF pressure between intracranial and spinal compartments. 26 In conclusion, our results demonstrated a significant relationship between ultrasound findings and CSF pressure. The measurement of combined ONSD and ODH may be a promising technique to detect elevated CSF pressure. A prospective multicenter clinical trial with a larger cohort is necessary to further evaluate the clinical significance of an enlarged ONSD and increased ODH in NSCLC-LM.