Combined imaging and surgical predictors of cardiophrenic lymph node metastasis in ovarian cancer patients undergoing upfront surgery

Highlights

• •A cardiophrenic lymph node with a short-axis diameter ≥ 7.7 mm may be suggestive of metastasis.
• •Diaphragmatic and intra-abdominal residual disease independently associated with cardiophrenic lymph node metastasis.
• •Combined assessment may provide useful guidance for surgical evaluation and planning in selected patients.

Abstract

Objectives

The aim of this study was to identify preoperative and intraoperative factors associated with cardiophrenic lymph node (CPLN) metastasis and to evaluate their potential role in surgical decision-making in patients undergoing upfront surgery for ovarian cancer. Because unsuspected CPLN involvement can compromise the ability to achieve no gross residual disease, early identification may guide surgical planning and selection of candidates for CPLN dissection.

Methods

From 2010 to 2023, electronic medical records of patients with upfront cytoreductive surgery including cardiophrenic lymph node (CPLN) dissection were reviewed and retrospectively analyzed.

Results

A total of 165 patients underwent upfront cytoreductive surgery, including CPLN dissection, and 100 patients (60.6 %) were confirmed CPLN metastasis. Patients with CPLN metastasis showed a larger short-axis diameter on CT than those without CPLN metastasis (7.95 mm and 6.8 mm, p < 0.0001, respectively) and the best predictive cutoff value for short-axis length was 7.7 mm. (sensitivity 0.600 and specificity 0.754) In multivariate analysis, seeding on the diaphragm was associated with CPLN metastasis (p < 0.0001), and intra-abdominal residual disease was associated with CPLN metastasis (p = 0.0029).

Conclusion

This study suggested preoperative and intraoperative factors associated with CPLN metastasis in ovarian cancer, including short-axis diameter, diaphragmatic involvement, and residual disease. While each factor has limitations as a standalone predictor, their combined assessment may provide useful guidance for surgical evaluation and planning in selected patients.

1. Introduction

Complete cytoreductive surgery is the most important determinant of survival in patients with advanced ovarian cancer. The presence of extra-abdominal metastasis, particularly to the CPLN, presents a surgical challenge, as it may compromise the ability to achieve no gross residual disease. CPLNs belong to parietal supradiaphragmatic lymph-node group and are located in the fat at the base of mediastinum within extrapleural space.(Agusti et al., 2024, Holloway et al., 1997) Enlarged CPLN is reported in 15–28 % of the ovarian cancer patients in CT imaging, depending on the radiologic definition applied (Holloway et al., 1997, Prader et al., 2016) CPLN metastasis is classified as stage IVB under the FIGO system and several studies have reported that enlarged CPLN may have a negative impact on progression-free survival and recurrence.(Larish et al., 2020, Kengsakul et al., 2021) Notably, Mert et al. reported that enlarged CPLN negatively impacted overall survival even in patients who achieved complete intra-abdominal cytoreduction. (Mert et al., 2018) However, CPLNs are located in the thoracic cavity, making direct visual assessment during surgery difficult. As a result, preoperative or intraoperative prediction of CPLN involvement is essential to support appropriate surgical planning and individualized treatment strategies.

Computed tomography (CT) is the primary preoperative imaging modality. Various cutoff values have been proposed, ranging from 5 to 10 mm. While a short-axis diameter of ≥ 10 mm has demonstrated high sensitivity in some studies, others reported better diagnostic performance using a ≥ 7 mm threshold. (Prader et al., 2016, Forstner et al., 2010, Eisenhauer et al., 2009, Kim et al., 2016) Nonetheless, these radiologic criteria remain limited by suboptimal sensitivity and specificity.

In particular, CPLN involvement has been correlated with a higher burden of upper abdominal disease.(Kengsakul et al., 2021, Luger et al., 2020) Since the extent of carcinomatosis directly affects the feasibility of optimal debulking, diaphragmatic involvement—frequently evaluated intraoperatively by liver mobilization (Rosati et al., 2024)—may serve as a potential surrogate marker for thoracic lymphatic spread. However, the diagnostic accuracy ^of such intraoperative findings, either alone or in combination with CT imaging, remains poorly defined.

Therefore, this study aimed to identify preoperative and intraoperative factors associated with CPLN metastasis and to explore their potential utility in guiding surgical decision-making for patients undergoing upfront cytoreductive surgery for ovarian cancer.

2. Methods

2.1. Patients and study design

From January 2010 to December 2023, we reviewed records of 165 patients with ovarian cancer who underwent upfront cytoreductive surgery, including CPLN dissection. We excluded patients with any history of systemic chemotherapy or radiotherapy, including those treated with neoadjuvant chemotherapy for ovarian cancer. Preoperative CT and serum analysis, as well as surgical and pathological reports, were reviewed for all patients.

Before treatment, all patients received physical examination and abdominopelvic and chest CT and were assessed to be clinically stage III or higher, before treatment. Images and treatment plans were reviewed at a multidisciplinary team meeting consisting of gynecologic oncologists, radiologists, pathologists, medical and radiation oncologists and nursing staff. Based on multidisciplinary team meeting consensus, CPLN dissection during upfront cytoreductive surgery was selected when an enlarged CPLN on preoperative CT was judged to have a high likelihood of metastasis and complete intra-abdominal cytoreduction was otherwise intended. Patients deemed not feasible for upfront surgery based on patients’ general condition and images underwent cytoreductive surgery after neoadjuvant chemotherapy.

All patients underwent upfront surgery, including salpingo-oophorectomy and other intra-abdominal cytoreduction. Further cytoreductive procedures were performed in consultation with colorectal, urologic, hepatobiliary and thoracic surgeons, led by a gynecologic oncologist. CPLN dissection was considered when preoperative CT suggested suspicious enlargement and complete intra-abdominal cytoreduction was planned. CPLN dissection was performed via transdiaphragmatic approach or video-assisted thoracoscopic surgery (VATS), and detailed surgical approaches were described in previous studies. (Lim et al., 2009, Yoo et al., 2013) CPLN dissection was primarily performed when preoperative CT showed lymph node enlargement and advanced-stage ovarian cancer was suspected.

The short-axis diameter of the CPLN was measured on the axial view of the preoperative CT scan at the level where the lymph node was most prominent, using standard PACS calipers. A gynecologic oncologist with 5 years’ surgical experience remeasured CPLN short-axis diameters on axial CT slices while blinded to all clinical and pathological information.

Diaphragmatic seeding or involvement was defined as macroscopic peritoneal implants or nodules on the diaphragmatic peritoneum visualized intraoperatively (often after liver mobilization) and confirmed pathologically. Falciform ligament involvement was defined as visible tumor deposits on the falciform ligament with pathological confirmation. Intra-abdominal residual disease was defined as any gross residual tumor remaining at the completion of cytoreductive surgery.

This retrospective study was approved by the Institutional Review Board of the National Cancer Center, Korea (NCC2024-0282), and informed consent was waived. Inpatient and outpatient medical records were collected.

2.2. Statistical analysis

SAS 9.4 version (SAS Institute Inc., Cary, NC, USA.) and R version 4.3.3 (R Foundation for Statistical Computing) were used for all statistical analyses. A p value less than 0.05 was regarded as significant in all analyses. Continuous and categorical data were presented as median (minimum – maximum) and percentages, respectively. The t-test was used to evaluate the difference in age between the two patient groups, and Wilcoxon rank-sum test was used for serum cancer antigen 125 (CA-125) levels and CPLN short-axis diameter. The Chi-squared test and Fisher’s exact test were used to compare the tumor invasion of the diaphragm, falciform ligament, abdominopelvic lymph nodes, and pleura between the two groups.

The optimal CPLN short-axis cutoff was determined using Receiver Operating Characteristic (ROC) curve, and the sensitivity and specificity were also calculated. To identify independent risk factors of CPLN metastasis, clinicopathological characteristics that appeared significant on univariable logistic regression analysis were entered into a multivariable logistic-regression model. Variables were selected using backward elimination method, leaving only significant variables in the multivariable analysis. Associations were presented with the odds ratio and corresponding 95 % confidence interval (CI) derived from the model.

3. Results

3.1. Clinical characteristics

Among the 165 total patients with CPLN dissection, 100 patients (61.6 %) had CPLN metastasis and 65 had no metastasis of CPLN. The characteristics of the two groups are presented in Table 1. Among 165 patients who underwent CPLN dissection, optimal intra-abdominal cytoreduction (<1 cm macroscopic residual disease) was achieved in 161/165 (97.6 %), with only 4 patients not reaching this threshold.

Table 1.

Clinical characteristics based on cardiophrenic lymph node metastasis status.

		no CPLN metastasis	CPLN metastasis	P-value
		N = 65	N = 100
Age				0.0679*
	Median (IQR)	52 (44–59)	55 (48–63)
Pathologic stage
	I	7 (10.8 %)
	II	5 (7.7 %)
	III	38 (58.5 %)
	IV	15 (23.1 %)	100 (100 %)
Grade				0.7606‡
	1	2 (3.2 %)	5 (5 %)
	2	11 (17.7 %)	14 (14 %)
	3	49 (79 %)	81 (81 %)
Histology				0.0060‡
	HGSOC	46 (70.8 %)	86 (86 %)
	Endometrioid	4 (6.2 %)	0 (0 %)
	Clear cell	1 (1.5 %)	6 (6 %)
	LGSOC	1 (1.5 %)	3 (3 %)
	Mucinous	1 (1.5 %)	0 (0 %)
	Carcinosarcoma	6 (9.2 %)	2 (2 %)
	Others**	6 (9.2 %)	3 (3 %)
Preoperative CA125, IU/ml				0.2465 #
	Median (IQR)	821 (496–1844)	1061.5 (60.9–2543.5)
CPLN short diameter on CT, mm				<0.0001 #
	Median (IQR)	6.8 (5.7–7.6)	7.95 (7–9.40)
Residual disease				<0.0001‡
	No gross residual disease	55 (84.6 %)	50 (50.0 %)
	Macroscopic < 1 cm	9 (13.9 %)	48 (48.0 %)
	Macroscopic ≥ 1 cm	2 (3.1 %)	2 (2.0 %)

*: T-test, #: Wilcoxon rank sum test, †: Chi-squared test, ‡: Fisher’s exact test.

** mixed type, non-epithelial tumor.

CPLN, cardiophrenic lymph node.

HGSOC, high grade serous ovarian cancer; LGSOC, low grade serous ovarian cancer.

The median age of the two group did not differ between groups. All 100 patients with metastatic CPLN were documented as stage IV. High-grade carcinoma predominated in both cohorts with relatively higher representation in the metastasis group, whereas histologies with grade 1 or 2 were less frequent in both groups. The tumor grade between two groups was not significantly different. High-grade serous ovarian cancer (HGSOC) was the most common histology in both groups, but the distribution of tumor histology differed between groups (p = 0.0060) The median preoperative serum CA-125 level was not significantly different between the groups. (1061.5 and 821, respectively) The mean CPLN size was 7.9 mm and the median was 7.5 mm. Overall, 25/165 patients (15.2 %) had CPLNs with ≥ 10 mm short-axis diameter; of these, 20/25 (80.0 %) demonstrated pathologic CPLN metastasis. Patients with CPLN metastasis had a larger median short-axis diameter of CPLN on preoperative CT imaging (7.95 mm) than those without CPLN metastasis. (6.8 mm, p < 0.0001) Postoperative residual disease was more frequent in patients with CPLN metastasis. (p < 0.0001) Among 165 patients, 4 patients underwent suboptimal cytoreduction (residual disease ≥ 1 cm). Their CPLN short-axis diameters on preoperative CT were 7.7, 7.8, 7.9, and 9.1 mm, respectively, and all four nodes were pathologically positive.

3.2. Analysis with the optimal cutoff for short-axis length

The ROC curve using the short-axis diameter of CPLN is shown in Fig. 1. The maximum Youden index was obtained with a cutoff value of 7.7 mm for the CPLN short-axis diameter on preoperative CT imaging. The sensitivity of this value is 0.600 and the specificity is 0.754.

Fig. 1.

ROC curve using the short-axis diameter of enlarged cardiophrenic lymph nodes. The sensitivity and specificity are 0.6 and 0.75, respectively. A cut-off value of 7.7 mm for the short-axis diameter is shown.

3.3. Other pathologically proven tumor involvement

Regions confirmed as sites of pathological tumor invasion are shown in Table 2. Among patients without CPLN metastasis, 37 patients (57 %) had tumor involvement in the diaphragmatic peritoneum, while among those with CPLN metastasis, 96 patients (96 %) had metastasis to the diaphragmatic peritoneum (p < 0.0001) Abdominopelvic lymph node metastasis did not differ between two groups, including pelvic and paraaortic lymph node metastasis subgroups. No significant difference was observed in liver metastasis including both capsule and parenchyma involvement. Pleural invasion was observed in only two patients with CPLN metastasis.

Table 2.

Pathologic proven tumor involvement.

	no CPLNmetastasis	CPLNmetastasis	P-value
	N = 65	N = 100
Diaphragmatic peritoneum	37 (57 %)	96 (96 %)	0<.0001†
Falciform ligament	11 (17 %)	43 (43 %)	0.0005†
Retroperitoneal lymph node	33 (51 %)	50 (50 %)	0.9231‡
Pelvic	30 (46 %)	45 (45 %)
Paraaortic	31 (48 %)	47 (47 %)
Liver	11 (17 %)	24 (24 %)	0.2772†
Capsule	8 (12 %)	20 (20 %)
Parenchyma	3 (5 %)	4 (4 %)
Pleura	0	2 (2 %)	>0.9999‡

†: Chi-squared test, ‡: Fisher’s exact test.

CPLN, cardiophrenic lymph node.

3.4. Factors associated with cardiophrenic lymph node metastasis

Logistic regression analysis is shown in Table 3. In multivariable logistic regression analysis, postoperative residual disease, CPLN short-axis diameter on preoperative CT imaging and diaphragm tumor involvement were identified as significant independent factors in predicting CPLN metastasis. (p = 0.0002 and p < 0.0001, respectively) When the CPLN short-axis diameter was ≥ 7.7 mm, the odds increased 4.462 times. Additionally, when diaphragmatic tumor involvement was present, the odds increased by 17.694 times compared to cases without diaphragm metastasis.

Table 3.

Logistic regression analysis to predict cardiophrenic lymph node metastasis.

		Univariable		Multivariable
		OR (95 %CI)	P-value	OR (95 %CI)	P-value
Age		1.029 (0.998–1.061)	0.0698
Histology
	HGSOC	1 (Ref)
	non-HGSOC	0.394 (0.181–0.858)	0.0189
CT CPLN short diameter
	<7.7	1 (Ref)		1 (Ref)
	≥7.7	4.594 (2.300–9.175)	0<.0001	4.067 (1.807–9.154)	0.0007
Residual disease
	microscopic	1 (Ref)		1 (Ref)
	macroscopic	5.500 (2.522–11.991)	<0.0001	3.772 (1.572–9.049)	0.0029
Tumor involvement
Diaphragm
	No	1 (Ref)		1 (Ref)
	Yes	18.162 (5.961–55.341)	0<.0001	14.72 (4.443–48.766)	0<.0001
Falciform ligament
	No	1 (Ref)
	Yes	3.703 (1.733–7.916)	0.0007
Liver	No	1 (Ref)
	Yes	1.550 (0.700–3.430)	0.2794

CPLN, cardiophrenic lymph node.

3.5. A patient case

A 66-year-old woman with ovarian cancer underwent CT before surgery, which showed an enlarged CPLN, as shown in Fig. 2A (arrow). Intraoperatively, seeding lesions were observed on the diaphragm, leading to a diaphragmatic peritonectomy. (Fig. 2B) In the final pathological results, both the diaphragmatic peritoneum and CPLN were confirmed to be metastatic.

Fig. 2.

A patient with CPLN metastasis (A) enlarged CPLN on preoperative CT (arrow) (B) Peritoneal seeding on diaphragm (arrow). CPLN, cardiophrenic lymph node

4. Discussion

Of 165 patients, 100 (61.6 %) had pathologically confirmed CPLN metastasis. The short-axis diameter on preoperative CT was significantly larger in the CPLN metastasis group, with a cutoff value of ≥ 7.7 mm showing moderate specificity (0.754) but limited sensitivity (0.600). Diaphragmatic metastasis and intra-abdominal residual disease were independently associated with CPLN metastasis (p < 0.0001 and p = 0.0029, respectively). While each of these factors has limited diagnostic value on its own, their combined evaluation may offer meaningful support for surgical assessment and decision-making in selected patients.

Although the short-axis diameter of pathologic CPLNs was significantly larger than that of benign nodes, a threshold of ≥ 7.7 mm was identified as a potential indicator of metastasis. This finding implies that preoperative CT can be used to evaluate CPLN status. However, the sensitivity of this criterion was relatively low. These results suggest that while CT-based size assessment may provide guidance for CPLN metastasis, it is insufficient as a standalone diagnostic tool. Additional clinical parameters or advanced imaging modalities are warranted to improve diagnostic accuracy.

Given the anatomical role of the diaphragm as a major pathway for lymphatic drainage from the peritoneal cavity, its involvement may serve as an indirect indicator of CPLN metastasis. Lymphatic flow proceeds from the peritoneum to the retrosternal and anterior diaphragmatic regions, then into the thoracic cavity via intermediate lymph nodes before ultimately draining into the great veins or the cisterna chyli. (Abu-Hijleh et al., 1995, Sharma et al., 2020) Accordingly, peritoneal tumor spread involving the diaphragm may facilitate metastatic seeding to CPLNs through this lymphatic route. Intraoperatively, lifting the liver allows for visual inspection of diaphragmatic carcinomatosis, which can provide a valuable surrogate marker for evaluating potential CPLN involvement. (Fig. 2). However, given its lack of specificity, diaphragmatic metastasis alone should not be considered a definitive indicator of thoracic lymph node spread.

From a surgical perspective, pathologically enlarged CPLNs were found to be significantly associated with postoperative residual disease, which is a critical prognostic factor in primary ovarian cancer. Consistent with these observations, pathologic CPLNs were identified in all patients whose postoperative residual disease exceeded 1 cm, suggesting that CPLN enlargement may track with the burden of intra-abdominal tumor and the risk of resulting suboptimal surgery, thereby informing preoperative thresholds and intraoperative decision-making. Tailored cytoreductive strategies may benefit from evaluating CPLN involvement, particularly in patients with high-risk features such as diaphragmatic metastasis and extensive residual tumor.

However, the decision to perform CPLN dissection should be carefully weighed, as the invasiveness of the procedure must be balanced against its potential benefits. Not all patients will require this approach, and surgical planning should be individualized to optimize patient outcomes.

Although postoperative outcomes were not systematically captured in our dataset, prior reports suggest a low rate of major morbidity following CPLN resection.(Prader et al., 2016, Yoo et al., 2013, Lopes et al., 2019, Garbi et al., 2017, Cowan et al., 2017) Pulmonary events are most commonly described, with pleural effusion reported in approximately 9–55 % and uncomplicated atelectasis in about 25 % of patients. (Yoo et al., 2013) However, attributing cardiothoracic events solely to CPLN resection is challenging in the context of extensive cytoreductive surgery, where concomitant upper-abdominal procedures—such as diaphragmatic peritoneal stripping, falciform ligament resection, and hepatic surgery—can independently precipitate pleural effusion and related pulmonary findings.

This study has several limitations. First, its retrospective design may have introduced selection bias and limited the ability to control for potential confounding variables. Second, we did not assess inter-observer variability or formal concordance between radiologists and gynecologic oncologists for CPLN short-axis measurement. Because our measurement relies on routine axial CT images and simple caliper tools, it is operationally feasible to incorporate the CPLN short-axis assessment into either radiology-led or surgeon-led workflows. Embedding the short-axis value in radiology reports may further streamline adoption and improve consistency across centers. Third, intraoperative gross assessment of peritoneal and diaphragmatic disease is inherently operator-dependent and subject to inter-observer variability, which may limit accuracy and reproducibility. Although we attempted to mitigate this by using a pragmatic, blinded short-axis caliper measurement on preoperative CT and by applying consistent intraoperative definitions, these approaches do not fully remove subjectivity.

Also, because CPLN dissection is not commonly performed in upfront surgery, the sample size was relatively small and we were not able to fully evaluate its effect on survival outcomes. Our findings should also be interpreted in light of the LION trial, which evaluated systematic retroperitoneal lymphadenectomy in patients with suspected advanced ovarian cancer and showed no survival benefit when lymph nodes were clinically and radiologically normal.(Harter et al., 2017) While informative, LION addresses a different anatomic compartment and clinical scenario than the supradiaphragmatic CPLNs assessed here. Prior studies have reported mixed findings on the survival impact of CPLN involvement and/or resection. Our results should therefore be interpreted as aiding operative assessment and selection rather than as definitive evidence of survival benefit. Further prospective studies with larger cohorts are warranted to discuss the prognostic and survival impact of CPLN involvement or CPLN dissection.

In conclusion, this study suggested preoperative and intraoperative factors associated with CPLN metastasis in ovarian cancer, including short-axis diameter, diaphragmatic involvement, and residual disease. While each factor has limitations as a standalone predictor, their combined assessment may provide useful guidance for surgical evaluation and planning in selected patients.

5. Ethics committee approval

This study was approved by the Ethics Committees of the National Cancer Center.

Funding sources

No funding was received for this research.

CRediT authorship contribution statement

Uisuk Kim: Writing – review & editing, Writing – original draft, Methodology, Data curation, Conceptualization. Ji Hyun Kim: Writing – review & editing, Supervision, Methodology, Conceptualization. Jae Won Song: Supervision, Data curation. Jin-Ho Choi: Supervision, Data curation. Sun Ho Kim: Data curation. Sang-Yoon Park: Writing – review & editing, Supervision. Myong Cheol Lim: Supervision, Methodology, Conceptualization.

Data availability

The datasets and scripts used in this study are available from the corresponding authors upon request.