Prognostic Value of Chest X-ray and CT Defined Large Mediastinal Adenopathy in High Risk Pediatric Hodgkin Lymphoma: A report from the Children’s Oncology Group Study AHOD0831

Abstract

Purpose/Objective(s):

We compared the prognostic value of chest radiograph (CXR)- and computed tomography (CT)-derived definition of large mediastinal adenopathy (LMA) in pediatric Hodgkin lymphoma (HL).

Materials/Methods:

143 patients treated for stage IIIB/IVB HL on COG AHOD0831 were included in this study. Six definitions of LMA were investigated: (i) mediastinal mass ratio on CXR (MR_CXR) >1/3; (ii) mediastinal mass ratio on CT (MR_CT) >1/3; (iii) mediastinal mass volume on CT (MV_CT) >200mL (iv) normalized mediastinal mass volume (MV_CT/thoracic diameter [TD]) >1mL/mm, (v) mediastinal mass diameter on CT (MD_CT) >10cm; and (vi) normalized mediastinal mass diameter (MD_CT/TD) >1/3.

Results:

Median age at diagnosis was 15.8 years (range: 5.2–21.3 years). In patients with a slow early response (SER) to chemotherapy, MV_CT>200mL, MD_CT>10cm and MD_CT/TD>1/3 were associated with worse relapse-free survival (RFS) on MVA, while MR_CXR>1/3, MR_CT>1/3 and MV_CV/TD>1 mL/mm trended towards worse RFS; MD_CT/TD was the most strongly prognostic for inferior RFS, with a hazard ratio of 6.41 for MD_CT/TD>1/3 vs. ≤1/3 on MVA (P=0.02).

Conclusion:

LMA according to MV_CT>200mL, MD_CT>10cm and MD_CT/TD>1/3 are associated with poor prognosis in advanced-stage HL patients with SER. The normalized mediastinal diameter, MD_CT/TD>1/3 appears to be the strongest predictor of inferior RFS.

Introduction

The presence of bulky disease in the mediastinum is known to be a poor prognostic factor in Hodgkin lymphoma.^1,2 Large mediastinal adenopathy (LMA) has historically been delineated using a standing posterior-anterior chest radiograph (CXR). Currently, there is variation across pediatric trial groups and staging systems on the definition of LMA. In Children’s Oncology Group (COG) studies, a mediastinal mass is considered LMA when mediastinal mass ratio (i.e., transverse mass diameter divided by the thoracic diameter [TD]) is equal to or greater than one-third. On the other hand, European Network Pediatric Hodgkin Lymphoma’s (EuroNet-PHL) defines LMA as mediastinal mass volume on CT (MV_CT) ≥200 mL, with MV_CT being half the product of the three largest perpendicular diameters³. In the Ann Arbor classification, a mass is considered “bulky” if its maximum dimension is greater or equal to 10 cm,⁴ and this definition of bulk was retained in the Lugano Classification.⁵

There have been few studies comparing LMA definitions in HL, especially in the pediatric population. While mediastinal mass ratio (MR)>1/3 on CXR is a long-established definition of LMA, it represents a limited evaluation of the mediastinal mass given that only its transverse dimension is accounted for. As CT has been a standard component of staging HL for several decades and enables 3-dimensional measurements, it may be a more practical and effective modality for defining LMA.

We sought to assess and compare the prognostic value of CXR and CT-derived LMA definitions in pediatric Hodgkin lymphoma. We hypothesized that the optimal bulk threshold for a mediastinal mass differs depending on the patient’s size. Thus, in addition to evaluating the prognostic significance of mediastinal mass volume (MV_CT)>200mL and maximum mediastinal diameter (MD_CT) >10cm, we examined these respective measurements normalized to thoracic diameter (MV_CT/TD>1 mL/mm and MD_CT/TD>1/3), which may allow for age-appropriate size-based normalization. Furthermore, we sought to determine if the current CXR based mediastinal mass ratio LMA definition could be translated to CT (MR_CT>1/3).

Materials & Methods

Patients

The study cohort consists of patients enrolled on the Children’s Oncology Group phase III clinical trial, AHOD0831.⁶ Eligible patients were diagnosed at less than 21 years of age between December 2009 and January 2011 with stage IIIB or stage IVB classical Hodgkin lymphoma. All patients were first treated with 2 cycles of induction chemotherapy with doxorubicin, bleomycin, vincristine, etoposide, prednisone and cyclophosphamide (ABVE-PC) administered every 21 days. Those who had a rapid early response [RER, no ¹⁸F-fluoro-2-deoxy-D-glucose positron emission tomography (PET) activity above mediastinal pool] were consolidated with 2 cycles of ABVE-PC. Those who had a slow early response (SER) were treated with 2 cycles of ifosfamide and vinorelbine plus 2 further cycles of ABVE-PC. Radiation therapy (RT) was administered to sites of initial bulk and/or SER. Bulky disease was defined as any one of (i) Large mediastinal adenopathy (LMA): mediastinal mass ratio (MR) > 1/3, meaning tumor diameter >1/3 the thoracic diameter at the level of the dome of the diaphragm on a posterior-anterior supine chest X-ray; (ii) A continuous aggregate of nodal tissue that measures >6 cm in the longest transverse diameter in the axial plane in any nodal area; or (iii) Macroscopic splenic nodules: focal defects in the spleen.

Radiologic Assessment

We investigated MR_CXR and five other definitions of LMA for a total of six parameters defined as follows:

1. Mediastinal mass ratio on CXR (MR_CXR), which is the largest tumor transverse diameter divided by the thoracic diameter at the level of the dome of the diaphragm on a posterior-anterior supine chest X-ray; this was measured prospectively and centrally reviewed as part of the AHOD0831 trial

2. Mediastinal mass ratio on CT (MR_CT), which is the widest transverse mediastinal tumor diameter divided by the thoracic diameter

3. Mediastinal mass volume on CT (MV_CT), which is half of the product of the 3 largest perpendicular diameters

4. Normalized mediastinal mass volume (MV_CT/TD), which is MV_CT divided by the thoracic diameter

5. Mediastinal mass diameter on CT (MD_CT), which is the longest diameter of the largest mediastinal mass in either the transaxial or cranial-caudal dimension

6. Normalized mediastinal mass diameter (MD_CT/TD), which is MD_CT divided by the thoracic diameter

K.M. and I.L. retrospectively conducted the necessary radiologic measurements that were not done by the COG AHOD0831 central review. Based on existing literature^3,7,8, LMA was defined as MR_CXR>1/3, MR_CT>1/3, MV_CT>200mL, MD_CT>10cm and MD_CT/TD>1/3.

As MV_CT/TD had not been previously examined, we chose an LMA definition of MV_CT/TD>1.0 mL/mm, since 1.0 mL/mm was a pragmatic threshold that was close to the median MV_CT/TD in our cohort (0.91 mL/mm).

Statistical Analysis

The observed proportionate agreement was calculated between the MR_CXR definition and each of the 5 CT definitions using Cohen’s kappa, with values ≤0 as indicating no agreement, 0.01–0.20 as none to slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.8 as substantial, and 0.81–1.00 as almost perfect agreement⁹. The Kaplan-Meier method was used to calculate relapse-free survival (RFS) and RFS between subgroups was compared using log-rank test. Multivariable analyses (MVA) evaluating the effect of LMA on RFS were performed using Cox proportional hazards model. Co-variables entered into the multivariable models were age at diagnosis, gender, race, stage, histology and response to chemotherapy. A stepwise selection process was used with p<0.1 to enter the model and p<0.15 to stay in in the model. P≤0.05 was considered statistically significant. All statistical analyses were performed with SAS (version 9.4, SAS institute Inc., Cary, NC).

Results

Of the 164 patients enrolled on AHOD0831, 143 had good quality CXRs and CT scans, comprising the study cohort. Patient, disease and treatment characteristics are shown in Table 1. Median age at diagnosis was 15.8 years (range: 5.2–21.3 years). Median follow up was 6.8 years.

Table 1:

Patient, disease and treatment characteristics (N=143)

Variable		N (%)
Age at Diagnosis (yr)	Mean±SD Median [min, max]	14.9±3.2 15.8 [5.2, 21.3]
Stage IV	No	58 (40.6%)
	Yes	85 (59.4%)
Gender	Male	89 (62.2%)
	Female	54 (37.8%)
Race	Black	24 (16.8%)
	Other	27 (18.9%)
	White	92 (64.3%)
Any radiotherapy	No	30 (21.0%)
	Yes	113 (79.0%)
Mediastinal radiotherapy	No	56 (39.2%)
	Yes	87 (60.8%)
Response to chemotherapy	RER	70 (49.0%)
	SER	73 (51.0%)
Histology	HL, NOS	24 (16.8%)
	HL, nodular sclerosis, NOS	95 (66.4%)
	Other*	24 (16.8%)

Abbreviations: SD=standard deviation; HL=Hodgkin lymphoma; NOS=not otherwise specified

^*Other includes HL, lymphocyte-rich; HL, mixed cellularity, NOS; HL, lymphocyte depletion, NOS; HL, nodular lymphocyte-predominant

Evaluation of the agreement between MR_CXR>1/3 and the 5 CT definitions is shown in Table 2. The rate of agreement was 88% for MR_CT>1/3, 79% for MV_CT>200mL, 81% for MV_CT/TD>1 mL/mm, 76% for MD_CT>10cm and 83% for MD_CT/TD>1/3. In the 34 patients who were 12 years and younger, the rate of agreement was 82% for MR_CT>1/3, 68% for MV_CT>1/3, 68% for MV_CT/TD>1 mL/mm, 59% for MD_CT>10cm and 85% for MD_CT/TD>1/3. 55% of the study cohort had LMA according to the MR_CXR>1/3 definition, compared to 66% according to the MR_CT>1/3 definition (P<0.001).

Table 2:

Comparison CXR vs. CT definitions of large mediastinal adenopathy

		Mediastinal mass ratio on CXR (MRCXR)
		No LMA	LMA
Mediastinal mass ratio on CT (MRCT)	No LMA	50	3
	LMA	14	76
	Rate of agreement	88.1%
Mediastinal mass volume on CT (MVCT)	No LMA	50	16
	LMA	14	63
	Rate of agreement	79.0%
Mediastinal mass volume on CT / TD (MVCT/TD)	No LMA	56	19
	LMA	8	60
	Rate of agreement	81.1%
Maximum mediastinal diameter on CT (MDCT)	No LMA	53	23
	LMA	11	56
	Rate of agreement	76.2%
Maximum mediastinal diameter on CT / TD (MDCT/TD)	No LMA	44	5
	LMA	20	74
	Rate of agreement	82.5%

Table 3 shows the 4-year RFS with vs. without LMA for patients with SER and RER to chemotherapy. For patients with RER, there was no significant difference between those with vs. without LMA based on any of the 6 definitions. For SER patients, those with MV_CT>200mL, MD_CT>10cm and MD_CT/TD>1/3 had significantly worse RFS compared to those without LMA; SER patients with LMA according to MR_CXR>1/3, MR_CT>1/3 and MV_CV/TD>1 mL/mm trended towards inferior RFS. Kaplan-Meir RFS curves for SER patients with versus without LMA are shown in Supplemental Figures 1A-F.

Table 3:

4-year RFS stratified by LMA vs. no LMA according to MR_CXR, MR_CT, MV_CT. MV_CT/TD and MD_CT, MD_CT/TD definitions

	RER patients (N=70)			SER patients (N=73)
	4-year RFS ± standard error (SE) (N; # of relapses)			4-year RFS ± standard error (SE) (N; # of relapses)
	LMA	No LMA	P	LMA	No LMA	P
Mediastinal mass ratio on CXR (MRCXR)	84.3±6.5% (N=32; 7)	80.5±6.6% (N=38; 5)	0.74	70.4±6.9% (N=47; 13)	91.1±6.1% (N=26; 3)	0.10
Mediastinal mass ratio on CT (MRCT)	80.2±6.3% (N=42; 8)	85.0±6.9% (N=28; 4)	0.66	71.0±6.8% (N=48; 13)	90.9±6.2% (N=25; 3)	0.12
Mediastinal mass volume on CT (MVCT)	76.4±7.3% (N=35; 8)	88.0±5.6% (N=35; 4)	0.23	66.6±7.7% (N=42; 13)	92.6±5.0% (N=31; 3)	0.020
Normalized mediastinal mass volume on CT (MVCT/TD)	73.2±8.2% (N=31; 8)	89.3±5.0% (N=39; 4)	0.10	67.5±8.1% (N=37; 11)	88.0±5.7% (N=36; 5)	0.083
Maximum mediastinal diameter on CT (MDCT)	78.0±7.3% (N=32; 7)	86.1±5.8% (N=38; 5)	0.37	62.4±8.7% (N=35; 12)	91.3±4.8% (N=38; 4)	0.0089
Normalized maximum mediastinal diameter (MDCT/TD)	79.3±6.2% (N=45; 9)	87.3±6.9% (N=25; 3)	0.45	69.6±6.8% (N=49; 14)	94.4±5.4% (N=24; 2)	0.043

Abbreviations: RFS=relapse-free survival; CT=computed tomography; CXR=chest x-ray; LMA=large mediastinal adenopathy; TD=internal thoracic diameter; HL=Hodgkin lymphoma; NOS=not otherwise specified

Final models for the MVAs evaluating the effect of LMA in SER patients are displayed in Table 4. As shown, LMA was significantly associated with RFS according to MV_CT>200mL, MD_CT>10cm and MD_CT/TD>1/3. There was a trend towards worse RFS in patients who had LMA according to MR_CXR>1/3, MR_CT>1/3 and MV_CV/TD>1 mL/mm. The LMA definition that most strongly correlated with inferior RFS on MVA was MD_CT/TD>1/3 with a hazard ratio 6.41, followed by MD_CT>10cm with a hazard ratio of 4.04.

Table 4:

Final multivariable analysis (MVA) models for EFS in SER patients^#

LMA definition	Variable	Parameter	SD	Hazard ratio	P-value
Mediastinal Mass ratio on CXR (MRCXR)	MRCXR+ vs. MRCXR−	1.02	0.64	2.78	0.11
Mediastinal Mass ratio on CT (MRCT)	MRCT+ vs. MRCT−	0.97	0.64	2.64	0.13
Mediastinal mass volume on CT (MVCT)	MVCT+ vs. MVCT−	1.66	0.67	5.27	0.014
	Black vs. white race	0.86	0.59	2.35	0.038
	Other vs. white race	−2.20	1.09	0.11
	HL NOS vs. HL, nodular sclerosis	1.87	0.67	6.54	0.013
	Other vs. HL, nodular sclerosis	−0.27	1.08	0.76
Normalized mediastinal mass volume on CT (MVCT/TD)	MVCT/TD+ vs. MVCT/TD−	0.90	0.54	2.47	0.094
Maximum mediastinal diameter on CT (MDCT)	MDCT+ vs. MDCT−	1.40	0.58	4.04	0.016
Normalized maximum mediastinal diameter on CT (MDCT/TD)	MDCT/TD+ vs. MDCT/TD−	1.86	0.79	6.42	0.018
	Black vs. white race	0.97	0.61	2.63	0.031
	Other vs. white race	−2.21	1.10	0.11
	HL NOS vs. HL, nodular sclerosis	2.06	0.68	7.87	0.0054
	Other vs. HL, nodular sclerosis	−0.51	1.07	0.6

^#MVAs adjusted for age at diagnosis, gender, race, stage, histology and response to chemotherapy. *Other includes HL, lymphocyte-rich; HL, mixed cellularity, NOS; HL, lymphocyte depletion, NOS; HL, nodular lymphocyte-predominant. Abbreviations: CT=computed tomography; CXR= chest x-ray; TD=internal thoracic diameter; HL=Hodgkin lymphoma; NOS=not otherwise specified

Discussion

To our knowledge, this is the largest study comparing CXR and CT-based LMA definitions in pediatric Hodgkin lymphoma. Our findings show substantial agreement between the conventional CXR mediastinal ratio definition and the 5 CT-based LMA definitions investigated. In patients who had a SER to chemotherapy, the presence of LMA according to each definition predicted for either a significantly inferior RFS, or a trend towards it, despite all of these patients receiving escalated treatment with ifosfamide and vinorelbine plus mediastinal RT.

Previous studies in adult Hodgkin lymphoma have compared mediastinal ratio (or slight variations thereof) on CXR vs. CT. Kriz et al. investigated mediastinal ratio on CXR vs. CT in 145 patients of the German Hodgkin Study Group, using the same definition as that used in the current study (i.e., extent of tumor in its largest extent / thoracic diameter at the height of diaphragm)⁷. With a threshold of MR>1/3 for both modalities, Kriz’s study found that 58% of mediastinal tumors were classified as large on CT whereas 65% were classified as large on CXR. Using a slightly different definition, Bradley et al. compared thoracic ratio between CXR and CT (widest mediastinal including the mass divided by the internal thoracic diameter at T5/6) in 95 adults. The CT-derived thoracic ratios were consistently slightly larger than the CXR-derived thoracic ratios by approximately 2%⁸. Similar to Bradley’s study, the CT-derived mediastinal ratios in our study cohort tended to be larger than the CXR-derived mediastinal ratios, with 63% of tumors classified as LMA on CT versus 55% on CXR. On univariable analysis and MVA, MR_CT was comparable to MR_CXR in trending towards increased relapse risk in SERs, suggesting that CT is a reasonable alternative to CXR in measuring MR.

Compared to MR, MV_CT is potentially a more comprehensive measurement of bulk, accounting for volumetric rather than only transverse tumor dimensions. MV_CT of ≥200 mL has thus been employed by the European Network for Paediatric Hodgkin Lymphoma (EuroNet-PHL) to define LMA. In the second interim analysis of the EuroNet-PHL-C1 trial, early-stage patients with a high erythrocyte sedimentation rate and/or MV_CT≥200 mL had a worse prognosis (event-free survival [EFS] approximately 80%) compared to early-stage patients without these risk factors (EFS 97%)³. Our study demonstrates that MV_CT>200 mL predicts for inferior RFS in advanced-stage HL as well. The predictive threshold for MV_CT appears independent of patient size, given that normalizing MV_CT to thoracic diameter did not strengthen the correlation with RFS.

A greater maximum diameter of mediastinal adenopathy has been shown to portend a worse prognosis in predominantly adult populations^8,10,11. North et al. found that mediastinal mass diameter >7.5 cm on CXR had a greater intrathoracic relapse rate¹⁰. Bradley et al. observed that CT diameter >10cm was associated with significant increase in disease progression, while thoracic ratio on CXR did not have the same prognostic relevance⁸. Our study shows that the prognostic significance of maximum mediastinal diameter applies similarly to pediatric patients, with its predictive value further strengthened when normalized to the child’s thoracic diameter to account for variations in the patient’s size. The importance of thoracic diameter normalization is especially evident in patients ≤12 years of age, as the rate of agreement with the conventional CXR definition is almost perfect (85%) for MD_CT/TD>1/3 versus only moderate (59%) for MD_CT>10cm.

Our study highlights that patients with SER and LMA represent a very high-risk group despite intensified chemotherapy and RT, suggesting that further treatment augmentation should be considered in these patients. Given the increased toxicities from augmented treatment, however, refinement of risk stratification by optimizing the definition of LMA is important. Our results show that CT-based definitions of LMA, especially normalized mediastinal diameter (MD_CT/TD)>1/3, are stronger predictors of relapse than the conventional CXR definition (MR_CXR>1/3). In the study’s sample size of 73 patients with SER, applying MD_CT/TD>1/3 instead of MR_CXR>1/3 to define LMA and dictate treatment augmentation might have prevented an additional relapse. Further refinement of CT-based LMA definitions and incorporation into the staging of pediatric HL would facilitate further improvement in outcomes. Thus, while replacing CXR is daunting given the extensive data that has already been collected with CXR, the sooner the transition to CT is made, the sooner the increased prognostic data offered by CT can be used to benefit pediatric HL patients. Furthermore, since MR _CXR>1/3 was established as an LMA definition with older treatment paradigms including larger RT fields, there is rationale to re-examine and re-define LMA in the context of modern treatments as used in AHOD0831 and beyond.

This study should be interpreted in the context of its strengths and limitations. Strengths include the prospectively collected clinical data, protocol-directed treatment, and large sample size for a pediatric-only cohort. While MV_CT>200 mL and MD_CT>10cm have been previously established as LMA definitions, our study is novel in investigating the normalization of MV_CT and MD_CT to thoracic diameter, accounting for the potential impact of age and patient size on bulk thresholds. Since all patients with SER received RT, the presence or absence of LMA was not confounded by the receipt of RT in this group; on the other hand, as bulk was the only indication for RT in RER patients, our assessment of the impact of LMA after a RER is limited. Nevertheless, the lack of RFS difference in RER patients with vs. without LMA suggests that mediastinal RT likely compensates for the negative prognostic effect of LMA. Other limitations of our study are the small number of events and the inclusion of only advanced-stage HL patients; although the findings are statistically significant, it would be worthwhile to validate them in larger groups of patients and in early-stage HL. Lastly, as the RT dose was 21 Gy for all patients in this study, it is unknown whether the prognostic value of the investigated LMA definitions would differ in the context of higher RT doses.

Conclusions

Advanced-stage HL patients with LMA and a SER to chemotherapy are at a high risk of recurrence despite receiving escalated chemotherapy plus RT; further augmentation of therapy may be warranted in these patients. Mediastinal ratio based on CT (MR_CT) is comparable to the conventional CXR based definition (MR_CXR), while the addition of cross-dimensional measurements improves the prognostication of CT-defined LMA. Of all definitions investigated, maximum mediastinal diameter on CT (MD_CT) was the strongest predictor of inferior RFS, which was further strengthened when normalized to thoracic diameter (MD_CT/TD). Given the superior prognostication of CT- compared to CXR-defined LMA, transitioning away from the conventional MR_CXR definition is likely justified in pediatric HL patients. Furthermore, as CT is necessary to determine the overall initial disease extent, using CT-based definitions of LMA has the practical advantage of omitting an additional staging CXR. Further studies would be beneficial to validate the study findings in children with early-stage HL and in those have a RER to chemotherapy.

Abbreviations

CXR

chest x-ray

CT

computed tomography

HL

Hodgkin lymphoma

MR

mediastinal mass ratio

MR_CXR

mediastinal mass ratio on chest x-ray

MR_CT

mediastinal mass ratio on computed tomography

MV_CT

mediastinal mass volume on computed tomography

TD

thoracic diameter

SER

slow early response

RER

rapid early response

RFS

relapse-free survival

EFS

event-free survival

COG

Children’s Oncology Group

EuroNet-PHL

European Network Pediatric Hodgkin Lymphoma’s

ABVE-PC

doxorubicin, bleomycin, vincristine, etoposide, prednisone and cyclophosphamide

PET

positron emission tomography

RT

radiation therapy

MVA

multivariable analysis