Radiographic patterns of intrathoracic disease in breast carcinoma: Prognostic implications

Clinical Radiology(1993) 48, 253-257

Radiographic Patterns of Intrathoracic Disease in Breast Carcinoma: Prognostic Implications M. L. G A W N E - C A I N , S. R. M A L T H O U S E , J. F. R E I D Y , R. D. RUBENS*, P. S M I T H * and W. G R E G O R Y *

Department of Radiology, Guy's Hospital and *ICRF Division of Clinical Oncology, UMDS Guy's Hospital, London Bridge, London The chest radiographic appearances of patients with intrathoracic metastases from breast cancer are variable and their relationship to patient outcome is unknown. This study aimed to classify and determine the frequency of various radiographic patterns and, in addition, to establish whether these patterns can be used to predict patient prognosis. A retrospective study was made of the patients of a major breast unit who over a 3 year period were found to have developed intrathoracic metastases. In each case, the earliest plain chest Xray showing signs of metastasis was reviewed and the frequency of the different radiographic patterns determined. Comparison was made between these patterns and patient mortality. The chest radiographs of 92 patients who developed intrathoracic metastases during a 3 year period were analysed. Pulmonary nodules were found in 66%, pleural effusions in 41% (bilateral in 18%), lymphadenopathy in 25%, lymphangitis in 18% and pleurally-based nodules in 11%. Median survival of the whole group was 13.5 months. Patients with bilateral, but not unilateral, effusions had a significantly poorer prognosis (median survival 3 months). Although the median survival of patients with lymphangitis seemed poor (5.5 months), this difference was not statistically significant. Multivariate analysis showed bilateral effusions to be the only independent predictor of worse outcome. We have found that of the plain radiographic patterns seen in intrathoracic metastasis from breast cancer, only bilateral pleural effusions can be used to predict a worse prognosis than that. associated with intrathoracic metastasis as a whole. Gawne-Cain, M.L., Malthouse, S.R., Reidy, J.F., Rubens, R.D., Smith, P. & Gregory, W. (1993). Clinical Radiology 48, 253-257. Radiographic Patterns of Intrathoracic Disease in Breast Carcinoma: Prognostic Implications

Accepted for Publication 5 March 1993

Intrathoracic metastases from breast carcinoma are a frequent complication of a c o m m o n disease. Some 5777% of patients dying with breast cancer have autopsy evidence of pulmonary metastasis, whereas 50-70% have liver metastasis [I] and 47-85% have bone metastasis [2]. Some patterns of metastasis carry a worse prognosis than others. In one series, for example, it was found that patients developing their first sign of metastatic disease in bone, had a median survival of 24 months compared with a median survival of 3 months for first metastasis in liver [3]. Intrathoracic metastasis is usually first identified on chest X-ray and at this stage can have various manifestations ranging from single or multiple pulmonary nodules to pleural effusions or lymphangitis carcinomatosis. Although certain patterns, such as iymphangitis, are believed to confer a worse prognosis, there is as yet no evidence to support this. Indeed, although the various radiographic patterns of intrathoracic metastasis have been well described [4,5], only one previous attempt has been made to determine the incidence of these different patterns of metastasis and the study did not concentrate on specific radiographic findings [6]. In this study we aimed to classify the initial radiographic appearances of patients with intrathoracic metaCorrespondence to: Dr M. L. Gawne-Cain, Department of Diagnostic Radiology, St George's Hospital, Blackshaw Road, London SW17 0QT.

stases from breast cancer, to determine the incidence of these patterns, and to assess which, if any, features might be used to predict patient prognosis.

P A T I E N T S AND M E T H O D S Patient Recruitment

A retrospective study was made of the patients of this Breast Unit who, over the 3 year period 1/1/87 to 31/12/89 were found to have developed intrathoracic metastases for the first time. These metastases were diagnosed on CXRs performed either as part of baseline investigations at initial presentation or following development o f chest symptoms. Patients were included who had already been found to have recurrence of disease at other sites. Index Film Selection

The single C X R corresponding most closely in time with the designated date of development of intrathoracic metastasis was used. This was compared if necessary with previous CXRs in the packet, but CXRs performed later than this date were not available to the radiologist scoring the index film. The aim was to mimic normal reporting conditions at the time of diagnosis of intrathoracic metastases.

254

CLINICAL RADIOLOGY

Table 1 - Descriptive groups used in scoring CXRs

Group (subgroup)

Notes used by radiologists

Pulmonary nodules

A lesion is not to be included as a nodule if it has already been included as lymphangitis, lymphadenopathy or pleural based nodules. The radiologist must be confident that there is pleural fluid. Uncertain blunting of the costophrenic angle is not enough for inclusion in this group. e.g. Definite upper mediastinal widening, hilar mass (but not if the lesion has already been scored as a pulmonary nodule). Either parenchymal nodularity (markings 1 5 m m with hazy border) or interstitial linear pattern. Septal lines. One or more of the nodules sited peripherally forming an obtuse angle with the chest wall. Apparently normal lung, hilum and pleura on one side. State other features not included above.

Pleural effusion (unilateral or bilateral) Lymphadenopathy Lymphangitis carcinomatosa Pleural-based nodules Normal side Other

Each C X R was scored for the presence/absence of each feature.

Scoring of the Index Film

Patients Excluded

Index CXRs were scored for presence or absence of several radiographic features (Table 1, Fig. 1.) In an initial pilot study, six radiologists independently reviewed 15 films and scored them. The results from this were used, where necessary, to improve definitions of the above groups (Table 1). The features of lymphangitis were taken from Janower and Blennerhassett [7]. In the main study index CXRs were scored independently by two radiologists ( M G C and SM) and the final score determined by agreement between them.

Of the 92 CXRs reviewed, four were excluded: one because the CXR was not of sufficient quality to be scored, and three because there was a likely alternative explanation for the radiographic abnormalities, e.g. heart failure. One film was included in the descriptive part of the study but not included in the clinical correlation because of incompleteness of clinical information on the database.

Clinical Correlation

Mortality Survival curves were constructed for each group according to the method of Kaplan and Meier [8] and median survival determined. Because a single patient could have more than one feature present on the same CXR, multivariate analysis was performed to determine to what extent each feature was contributing to the outcome.

Patterns of Metastasis The overall incidence of the different radiographic features is shown in Table 2. Pulmonary nodules and pleural effusions were the most common finding (occurring in 66% and 41% respectively). In 41 patients (46%) abnormalities were isolated (abnormalities in one group only, e.g. pulmonary nodules but no other features on CXR), whereas the remaining 54% had multiple CXR abnormalities (Table 2). Pulmonary nodules were the most likely to be an isolated abnormality (32% of the whole group, 48% of all patients with pulmonary nodules).

Disease Characteristics

Mortality

The age, tumour size, node status and histology at presentation were determined. In addition, the relationship between bilateral pleural effusions and metastases to extrathoracic sites was investigated.

Median survival of the whole group and of the various subgroups (from time o f first development o f intrathoracic metastases) is shown in Table 3, and the survival curves for subgroups of patients with effusions and lymphangitis in Fig. 2. Multivariate analysis showed bilateral pleural effusion to be the only independent predictor of outcome (P = 0.002).

Statistical Methods Survival curves were constructed as described above. The probability of significant differences between the survival curves of different pairs of groups (in most cases, between patients with and patients without a particular CXR finding) was determined using the log rank test; a P value of less than 0.05 was considered significant. Multivariate analysis was performed using Cox's proportional hazards model [9].

RESULTS CXRs of 92 patients were reviewed. One hundred and twenty patients were recorded as having developed pulmonary metastases from breast cancer between 1/ 1/87 and 31 / 12/89. The CXRs of 28 patients were unavailable, generally because of patient follow-up at other hospitals.

Clinical Correlation Age range, tumour size, nodal status at time of diagnosis and histological type are shown in Table 4. Details of clinical features at presentation were unknown for 22-39% of the cases, generally because these cases were initially diagnosed elsewhere. Because of this, numbers were too small for meaningful comparison to be made with the CXR patterns ofintrathoracic metastasis. No significant difference was found between the distribution of extrathoracic metastases in patients with or without bilateral effusions. Liver metastases were present in one (7%) of the patients with bilateral pleural effusions, compared with seven (10%) of all other "patients (P=0.98). Bone metastases were present in four (25%) and thirty (44%) respectively (P= 0.32).

INTRATHORACIC DISEASE IN BREAST CANCER

255

(a)

(b)

(c) Fig. 1 Examples of CXR scoring. (a) Multiple pulmonary nodules. At least one nodule is cavitated, which is unusual. (b) Pleural-based nodules on the right with apparently normal left lung. (c) Bilateral lymphangitis carcinomatosa with bilateral pleural effusion.

Table 2 - Results of C X R scoring (n = 88)

Group

Nodules Effusions Lymphadenopathy Lymphangitis Pleural nodules Normal side Other

All patients > 5 nodules 1-5 nodules All Unilateral Bilateral

Bony mets

No. (%) of patients (n = 88)

No. (% of total) of patients with isolated abnormality

58 (66) 34 (39) 24 (27) 36 (41) 20 (23) 16 (18) 22 (25) 16 (18) 10 (11) 20 (23) 17 (19)

28 (32) 9 (10) 3 (3) 6 (7) 1 (1) 3 (3) 0 (0)

256

CLINICAL RADIOLOGY

Table 3 - M e d i a n survival o f patients according to C X R appearances

Group

Median survival (months) (significance (P ) )

Nodules

All patients > 5 nodules 1-5 nodules Unilateral Bilateral

Effusions

17.4 16.5 (0.092) 23 (0.092) 15 3 (0.002) 10 (0.869) 4 (0.269)

Lyrnphadenopathy Lymphangitis The median survival of the whole group was 13.5 months. Table 4 - Clinical features o f patients studied (n = 88)

Mean age Mean tumour size Node status positive for metastases at diagnosis Histology

100

Ductal carcinoma grade I Ductal carcinoma grade II Ductal carcinoma grade III Ductal carcinoma grade unknown Lobular carcinoma Medullary carcinoma Other

(a)

CHI = 12.29 ~~~~1~

80

r

53 years 3.9 cm 40

,

P= 0.0022

:~_

=~ o~ >

E

60

40 20

[~

~

, N=20

% I

1

, B'ilat N ': 16 I I

2

I

I

3 4 Time (years)

100

' N~ I

I

5

6

CpH_I= 0.9076 - 0.3407

9

(b)

80 Q

~ 60 ID

9~ "5 E

40

-~ 2 0 -

~ ~~ Absent N = 73 I' [ 1

I 2

I 3

I 4

I 5

,

Present

N = 14 I 6

Time (years) Fig. 2-Cumulative survival curves. (a) Patients with unilateral, bilateral or no pleural effusion. (b) Patients with or without lymphangitis carcinomatosa. Patients with bilateral pleural effusions had had a shorter median disease free interval since the original diagnosis (1.3 years) than those without (2.5 years) but this was not statistically significant ( P = 0 . 0 9 ) . The time from original cancer diagnosis to intrathoracic spread was similar in both groups (3.9 years with bilateral effusions, 3.5 years without).

I

(Range 24-78 years) (Range 0-12cm) (Known for 67 patients only (Known for 64 patients only

30 22 6 2 1

2

DISCUSSION We have shown that it is possible to classify the C X R appearances o f intrathoracic metastases f r o m breast cancer into five b r o a d groups. In this retrospective study, p u l m o n a r y nodules were the c o m m o n e s t manifestation (66%). Pleural effusions were f o u n d in 41% and were bilateral in 18%. Lymphangitis carcinomatosis was a less frequent finding (18%). Bilateral (but n o t unilateral) pleur-a| effusion was the 0 ~ y independent predictor o f a worse prognosis c o m p a r e d with the whole g r o u p o f patients with intrathoracic involvement in breast cancer. A l t h o u g h it is recognized that C T detects m o r e pulm o n a r y nodules than C X R [10-12], in clinical practice C T screening o f the thorax is undertaken in relatively few cases o f malignancy, Following generally accepted practice C X R s only were available rather than c o m p u t e d t o m o g r a p h y (CT) in this study because it was retrospective and we aimed to provide information which would be relevant in our normal clinical situation [13, 14]. Diagnosis o f metastasis was made on C X R appearance in the context o f proven primary breast carcinoma. This again mirrors the normal clinical situation where histological p r o o f o f metastasis is not routinely obtained unless definitive treatment such as resection is planned. Those cases for which there was a likely alternative interpretation o f the C X R appearances were excluded f r o m the study. The distribution o f intrathoracic metastases in this study is in keeping with the findings o f K a m b y et al. [6] who f o u n d p u l m o n a r y metastases in 71%, 'pleural recurrence' in 34% and 'mediastinal recurrence' in 20%. The study described was o f patients at the time o f first ever metastasis, unlike this study which included all newly diagnosed intrathoracic spread regardless o f earlier metastasis to other sites. In a previous study f r o m this unit [15], 10% o f 99 patients with malignant effusion f r o m breast cancer had bilateral effusions; in the current study 16 (44%) o f 36 effusions were bilateral. This m a y be'because the earlier study relied on the patient notes rather than reviewing the C X R s themselves. A p r o p o r t i o n o f the

INTRATHORACIC DISEASEIN BREAST CANCER small effusions detected in the c u r r e n t s t u d y m a y n o t h a v e been c o n s i d e r e d clinically significant in the earlier study, a n d therefore m a y n o t have been included. Bilateral pleural effusion was the o n l y i n d e p e n d e n t p r e d i c t o r o f p o o r e r p r o g n o s i s ( m e d i a n survival 3 m o n t h s , P = 0.002) whereas patients with u n i l a t e r a l pleural effusion h a d a m e d i a n survival o f 15 m o n t h s , n o t significantly different f r o m t h a t o f the whole g r o u p studied. W e h a d expected to find a worse p r o g n o s i s a s s o c i a t e d with l y m p h a n g i t i s c a r c i n o m a t o s i s but a l t h o u g h the m e d i a n survival for this g r o u p was low (5.5 m o n t h s ) , it was n o t statistically significant ( P = 0 . 3 4 ) ; s o m e p a t i e n t s in this g r o u p r e m a i n e d alive u p to 4 years later. A larger g r o u p o f p a t i e n t s m i g h t have shown the s h o r t e r life e x p e c t a n c y to be significant. It is n o t a b l e that the worse p r o g n o s i s a s s o c i a t e d with bilateral pleural effusions was highly significant ( P = 0 . 0 0 2 ) , a l t h o u g h b i l a t e r a l effusions o c c u r r e d at the same low frequency as l y m p h a n g i t i s (18%). T h e m e d i a n survival o f p a t i e n t s with u n i l a t e r a l effusions (15 m o n t h s ) is similar to t h a t in the earlier s t u d y f r o m this unit where the m e d i a n survival f r o m first d e t e c t i o n o f effusion was 15.7 m o n t h s [15]. W h y s h o u l d bilateral effusion have a p o o r prognosis? A pleural effusion in this s t u d y c o u l d be free pleural fluid o f a n y c o m p o s i t i o n . In the c o n t e x t o f m a l i g n a n c y , this fluid m a y be the result o f i n t r a t h o r a c i c m e t a s t a s e s (e.g. via l y m p h a t i c o b s t r u c t i o n o r a direct result o f p l e u r a l deposits) o r o f generalized m e t a s t a t i c disease (e.g. h y p o p r o t e i n a e m i a s e c o n d a r y to liver deposits). S o m e effusions m a y n o t be related directly to the m a l i g n a n c y , for e x a m p l e in h e a r t failure o r infection. P a t i e n t s w h o s e C X R signs h a d an o b v i o u s i n t e r p r e t a t i o n u n r e l a t e d to malign a n c y were excluded f r o m the study. A n effusion on C X R therefore is a non-specific indic a t o r o f disease. The fact t h a t bilateral pleural effusion is m o r e likely to be an isolated C X R finding (37% o f p a t i e n t s with bilateral effusions) t h a n unilateral effusion (15%) m i g h t be because it is m o r e often a s s o c i a t e d with e x t r a t h o r a c i c p a t h o l o g y . T h e worse p r o g n o s i s c o u l d then be related to this n o n - l o c a l p a t h o l o g y . A n a t t e m p t was m a d e to s u p p o r t this suggestion by d e t e r m i n i n g the incidence o f e x t r a t h o r a c i c m e t a s t a s e s in p a t i e n t s with b i l a t e r a l effusions, but it was n o t f o u n d to be significantly different f r o m the incidence in the o t h e r p a t i e n t s with chest metastases. A s h o r t e r m e d i a n disease-free interval in p a t i e n t s with bilateral effusions m i g h t suggest that effusions are associated with m o r e aggressive disease in

257

general, b u t this was n o t statistically significant with the small n u m b e r s in this study. In conclusion, we have s h o w n in this series that the presence o f b i l a t e r a l p l e u r a l effusions on the C X R o f a p a t i e n t with m e t a s t a t i c b r e a s t cancer confers a worse p r o g n o s i s t h a n o t h e r regularly seen C X R a b n o r m a l i t i e s . This a s s o c i a t i o n with p o o r p r o g n o s i s is m o r e m a r k e d t h a n t h a t o f l y m p h a n g i t i s c a r c i n o m a t o s i s which has been widely believed to be a m o r e sinister finding.

REFERENCES

1 Di Pietro S, Bertario L, Carta G, Re A. An analysis of 800 breast cancer patients relapsed after radical mastectomy. Tumori 1976;62:99-101. 2 Galasko CSB. The anatomy and pathways of skeletal metastases. In: Weiss L & Gilbert HA, eds. Bone metastasis. Boston, MA: GK Hall, 1981:49. 3 Coleman RE, Rubens RD. The clinical course of bone metastases from breast cancer. British Journal of Cancer 1987;55:61 66. 4 Crow J, Slavin G, Kreel L. Pulmonary metastasis: a pathologic and radiologic study. Cancer 1981;47:2595 2602. 5 Libshitz HI, North LB. Pulmonary metastases. Radiologic Clinics of North America 1982;20:437 451. 6 Kamby C, Vejborg I, Kristensen B, Olson LO, Mouridsen HT. Metastatic patterns in recurrent breast cancer with special reference to intrathoracic recurrences. Cancer 1988;62:2226-2233. 7 Janower ML, Blennerhassett JB. Lymphangitic spread of metastatic cancer to the lung. A radiographic pathologic correlation. Radiology 1971;101:267 273. 8 Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. American Statistical Association Journal 1958;53:457481. 9 Cox DR. Regression models and life tables. Journal of the Royal Statistical Society 1972;34:187-202. 10 Schaner EG, Chang AE, Doppman JL, Conkle DM, Flye MW, Rosenberg SA. Comparison of computed and conventional whole lung tomography in detecting pulmonary nodules: a prospective radiologic-pathologic study. American Journal of Roentgenology 1978;131:51-54. 11 Peuchot M, Libshitz HI. Pulmonary metastatic disease: radiologicsurgical correlation. Radiology 1987;164:719-722. 12 Chalmers N, Best JJK. The significance of pulmonary nodules detected by CT but not by chest radiography in tumour staging. Clinical Radiology 1991;44:410-412. 13 Dershaw DD, Osborne M. Imaging techniques in breast cancer. Seminars in Surgical Oncology 1989;5(2):82-939 14 Armstrong P. Pulmonary neoplasms. In: Grainger RG & Allison D J, eds. Diagnostic Radiology. Edinburgh: Churchill Livingstone, 1992:271-293. 15 Fentiman IS, Millis R, Sexton S, Hayward JL. Pleural effusion in breast cancer: a review of 105 cases. Cancer 1981;47:2087 2092.