Persistent mediastinal mass is not indicative of recurrence after chemotherapy only in paediatric Hodgkin's disease


H. van den Berg, Emma Children Hospital AMC, Academic Medical Centre, University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, The Netherlands. E-mail: h.vandenberg@


Most patients with Hodgkin's disease are treated with chemotherapy in conjunction with radiotherapy, but at the end of treatment a residual mass is often present. After combined therapy, it has been assumed that no additional treatment is needed. However, for children treated without radiotherapy, no data exist on the relevance of a residual mediastinal mass to risk of relapse. We report on the findings of follow-up thorax radiographs of a group of 27 children with initial mediastinal involvement, who were treated with chemotherapy only. We conclude that the regression rate of the mediastinal mass was not related to a later recurrence. Regression after chemotherapy without radiotherapy is probably slower than after combined therapy. We consider chest radiograph examinations to be appropriate for the follow-up of tumour regression. When the data were compared with a group of children with Hodgkin's disease without mediastinal involvement, we found that survival was not related to initial mediastinal involvement.

Most patients with Hodgkin's disease are treated with chemotherapy in conjunction with radiotherapy. In these patients, a residual mediastinal mass may be present for many months, the clinical significance of which is debatable. No data on the regression during and after chemotherapy of a mediastinal Hodgkin's lymphoma exist for those children treated with chemotherapy alone. We report the findings of follow-up thorax radiographs of 27 children with initial mediastinal involvement who were treated with chemotherapy only.

Materials and methods

All children treated for Hodgkin's disease from 1975 onwards in the Emma Children Hospital were considered for this study. Between 1975 and 1984, children with Hodgkin's disease presenting with lymph nodes of < 4 cm were given chemotherapy only. In children with nodes of > 4 cm, additional involved field radiotherapy was given. After 1984, all children, irrespective of stage and lymph node diameter, were treated with chemotherapy only (Behrendt et al, 1987, 1996; Van den Berg et al, 1997a, b.)

Those children (18 boys and nine girls) with Hodgkin's disease having mediastinal involvement and treated without radiotherapy between July 1975 and January 1994 were included in this study. Four children were treated with six mechlorethamine–vincristine–procarbazine–prednisone (MOPP) courses, nine children with four doxorubicin–bleomycin–vinblastine–decarbazine (ABVD) courses and 14 children were treated with three ABVD/MOPP courses. To examine survival in relation to the presence of a mediastinal mass at diagnosis, all 27 patients without initial mediastinal involvement and not treated with radiotherapy in the same period were used as controls. Of these, 14 had received MOPP, seven had received ABVD and six had received ABVD/MOPP.

Pretreatment evaluation consisted of a detailed patient history, physical examination, histological examination of an involved lymph node, complete blood count, liver function tests, serum uric acid level, chest radiograph, abdominal ultrasound examination and bone marrow aspiration. No splenectomies were performed. Staging was carried out according to the Ann Arbor recommendations.

Radiographs of the thorax were made on a regular basis during and after treatment. In the first year, the intervals were every 4 weeks; in the second year, every 6 weeks; in the third year, every 3 months; in the fourth year, every 4 months; in the fifth year, every 6 months. The radiographs of the chest were reviewed by two examiners. Scoring was as follows: (1) normal findings with good visualization of mediastinal paralines; (2) normal findings, without good visualization of paralines; (3) changes compatible with mediastinal localization of Hodgkin's disease; (4) overt mediastinal localizations of Hodgkin's disease, i.e. maximal mediastinal diameter > 1/3 of the thoracic diameter at Th5.

For statistical analysis of differences in distribution of characteristics, t-tests were applied. For correlations, the Pearson test and Spearman test were used. For differences in survival, a log-rank test was used.


Twenty-seven children with mediastinal involvement were included in the study. Age at diagnosis ranged from 4·3 to 16·3 years (median 13·6 years). Staging was as follows: six children, stage I; nine children, stage II; eight children, stage III; four children, stage IV. Histology was as follows: 23 nodular sclerosis, three mixed cellularity, one lymphocyte predominance.

Twenty-one children remain in continuous complete remission, follow-up time ranges from 11 to 256 months; median follow-up is 117 months. Six children experienced a relapse 11–67 months after treatment start; median time of occurrence of a relapse was 18 months. Data of the children who relapsed are given in Table I.

Examination of the radiographs showed that, at treatment cessation, 4 out of 27 cases had on radiographs non-massive localization compatible with Hodgkin's disease (score 3), in three cases minimal changes not indicative of Hodgkin's disease (score 2) were seen. In the remaining cases, regression was noted, but the radiographs were still scored as 4. None of the patients attained a score 1 during the treatment period.

All 21 patients in continuous complete remission ultimately achieved a score of 2. In 10 cases, a score of 1 was obtained (Table II). In these patients, no relapses were seen. In all relapse cases, a score of 3 was obtained, however in two children (cases 4 and 6) this was only after treatment of a relapse. Ultimately, a score of 2 was reached in all cases.

The interval between start of therapy and findings on radiographic examination did not correlate with clinical findings such as stage of the disease and histology subtype. Also, no differences between patients experiencing a relapse and children remaining in continuous complete remission were detected (t-tests P > 0·05). The median time to reach a score of 3 or 2 was even less in relapsed patients (Figs 1 and 2). However, the time interval required to obtain scores of 3 and 2, respectively, were highly correlated with each other (Spearman test, ρ 0·785, P < 0·001; Fig 3).

Figure 1.

Figure 1.

Interval in months after diagnosis to obtain a score of 2. Data of the patients are shown according to the occurrence of a relapse later (N, no; Y, yes).

Figure 2.

Figure 2.

Interval in months after diagnosis to obtain a score of 3. Data of the patients are shown according to the occurrence of a relapse later (N, no; Y, yes).

Figure 3.

Figure 3.

Correlation of the time intervals (months) to obtain scores of 2 and 3 (N, no; Y, yes).

Two of the 27 control cases without initial mediastinal involvement experienced a relapse. As a result, there was no difference in relapse rate with respect to the presence or absence of an initial mediastinal involvement (P = 0·25).


In most centres, Hodgkin's disease is treated with chemotherapy in conjunction with radiotherapy. According to the protocol, either mantle field or (nowadays more often) involved field irradiation is given. After finishing therapy, non-resolving mediastinal masses raise important clinical questions. A residual mass may indicate either persistent active disease, which could give rise to a relapse at a later date, or may be due to necrotic and fibrotic tissue. Additional treatment of such a non-malignant mass would mean that the patient receives unnecessary additional treatment, which could lead to serious late sequelae. These residual masses are, however, observed in many cases. After 11 months, 86% of the patients had a normal mediastinum (North et al, 1987). However, Orlandi et al (1990) reported that this is only in 38% of patients at the moment of treatment cessation. In their study, nine out of 102 patients had minimal anomalies and 30 out of 102 still had measurable changes (Orlandi et al, 1990). An even higher level of 64% of cases with a residual mass at the end of therapy was reported by Radford et al (1989). In 1 year, 59% of these cases showed regression of mediastinal remnants (Radford et al, 1989). However, at present, no data exist on the presence and significance of a residual mass after treatment cessation in children treated with chemotherapy only.

A second problem is how to define the extent of the mediastinal mass. There are at present no standard or ideal definitions available to measure regression (Hopper et al, 1991). The role of additional investigations in cases of a residual mediastinal mass is limited. If computerized tomography (CT) scanning is used, the incidence of patients with residual masses increases substantially, e.g. from 38% to 70% (Orlandi et al, 1990). Studies correlating CT scanning and chest radiographs have shown good concordances between both modalities (Brisse et al, 1998; Bradley et al, 1999). Movsas et al (1998), having followed patients receiving radiotherapy with weekly port films, showed that cumulative percentage regression was a statistically significant factor for in-field recurrence. The use of other diagnostic methods presents other dilemmas. The role of gallium scanning is limited because of the low sensitivity threshold, estimated at 1 cm, and because of the possibility of false positives due to physiological thymic uptake at the end of treatment (estimated frequency up to 43%) (Weiner et al, 1991). A combination of CT scanning and gallium uptake studies gives equivalent results to magnetic resonance imaging (MRI) studies in staging the patient. At restaging, gallium scanning has been reported to be superior to CT (Devizzi et al, 1997).

Confirmation of the nature of the lesion by resection of the residual mass has been performed only in a few cases. Fibrotic and necrotic material with signs of inflammation but without tumour involvement have been reported (Durkin & Durant, 1997). Brisse et al (1998) also found no tumour cells on histological examinations of the residual mass in nine cases; they concluded that in such instances of residual mass additional investigations are not needed and frequent chest radiographs are sufficient for follow-up. As a result, it is generally accepted that after combined treatment a stable residual mass is not correlated with a relapse and no additional therapy is required (North et al, 1987; Radford et al, 1989; Orlandi et al, 1990; Zinzani et al, 1992; Longo et al, 1997; Brisse et al, 1998). The presence of a residual mass after combined therapy has, however, in other studies been linked to a higher rate of relapse (Colonna et al, 1996). In addition, North et al (1987) reported that the intrathoracic relapse rate did not correlate with the speed of regression of the masses, but relapse occurred more than twice as often in patients with residual mediastinal widening. However, Orlandi et al (1990) suggested that only in those cases in which the patient was treated with chemotherapy alone a residual mass might be correlated with relapse, but they could not show statistical significance. The present findings are in contrast to such an assumption.

In several studies in adults, the prognosis of Hodgkin's disease was correlated with stage and presence of a mediastinal mass at diagnosis (Hutchinson et al, 1998). Radford et al (1989) found a higher relapse rate in 34 cases with a mediastinal localization who had been treated without radiotherapy. However, the policy of giving irradiation to those cases with mediastinal residual mass at the end of treatment would result in irradiation of 86% patients (Horning et al, 1996).

As indicated, the role of additional radiotherapy is debatable but seems to depend on differences in treatment regimens (Zinzani et al, 1992). For children, the prognosis in stages III and IV Hodgkin's disease does not differ significantly between those receiving MOPP courses and those treated with extended field irradiation (Hutchinson et al, 1998). There are, however, reports that using less alkylating agents the ultimate outcome is related to the presence of a bulky mass in the mediastinum at diagnosis (Ekert et al, 1993). North et al (1987) noted no difference in complete regression rate whether additional chemotherapy was administered or not. However, in their study, the size of the mediastinal mass at diagnosis was significantly related to the occurrence of a relapse.

Based on our data, we can make the following important conclusions. As the time points at which scores of 3 and 2 were obtained was the same for patients with a relapse and for those staying in complete remission, we conclude that the rate at which regression of a mediastinal mass occurs is not correlated with relapse. When we compare our data with that in the literature, it seems that radiotherapy induces a more rapid disappearance of the mediastinal mass. In our patients, prognosis was not linked statistically to mediastinal involvement at diagnosis, this is in contrast with the findings of Ekert et al (1993) using other chemotherapy. Our data confirm that a residual mass can be effectively evaluated by frequent conventional chest radiographs, thus making CT scanning unnecessary.