773

Periostitis in HypertrophicOsteoa rth ropathy:Relationship to Disease Duration

Carlos J. Pineda1Manuel Martinez-Lavin1

Julio E. Goobar�David J. Sartoris3

Paul Clopton3Donald ResniCk3

Received September 4, 1986; accepted afterrevision November 7, 1986.

This work was supported in part by VA Grant#SA 306 and Astra-Syntex Scandinavia AB, Sod-ertalje, Sweden.

I Department of Rheumatology, Instituto Na-

cional de Cardiologia. lgnacio Chave, Mexico 22.Mexico City.

2 Department of Rheumatology, Hospital for

Rheumatic Disease, 5-831 01 Osterund, Sweden.

3 Department of Radiology, veterans Administra-tion Medical Center and University of CalifomiaMedical Center, San Diego, CA 92161 . Addressreprint requests to D. Resnick, Dept. of Radiology,VA Medical Center, 3350 La Jolla village Dr., SanDiego, CA 92161.

AJR 148:773-778, April 19870361 -803X/87/1 484-0773© American Roentgen Ray Society

The relationship of periostitis to disease duration in primary hypertrophic osteoar-thropathy and the association of periostitis with cardiopulmonary disorders (secondarytype) were studied in order to define distinguishing features between the two. Radio-

graphic skeletal surveys were performed in 24 patients with hypertrophic osteoarthrop-athy to analyze pattern (single layer, multilayered, irregular) and site of involvement(diaphysis, metaphysis, epiphysis). The six patients with primary hypertrophic osteoar-thropathy and the 1 1 patients with cyanotic congenital heart disease had thicker, morewidespread penostitis involving the diaphysis, metaphysis, and epiphysis, in contrastto abnormalities in the seven patients with hypertrophic osteoarthropathy secondary tocarcinoma of the lung Average cortical bone widths as determined by radiogrammetricmeasurement of the second metacarpals were significantly greater for the patients withprimary hypertrophic osteoarthropathy (8.9 ± 6.0 mm) and cyanotic congenital heartdisease (8.5 ± 6.4 mm) as compared with the patients with bronchogenic carcinoma(6.0 ± 3.9 mm).

Correlation of radiographic patterns with duration of disease confirms that thicker,more extensive alterations are indicative of long-standing disease. The periostitis ofhypertrophic osteoarthropathy is therefore not dependent on the primary or secondarynature of the disease but principally on its duration.

Penostitis is a well-known radiographic feature of hypertrophic osteoarthropathy(HOA) and is frequently associated with digital clubbing, periarticular demineraliza-tion, and joint swelling or effusion. The morphology, thickness, and distribution ofperiosteal new-bone formation in patients with HOA have not been quantitativelystudied or well characterized. The literature contains conflicting observations onradiographic features that allow differentiation between primary and secondaryHOA. Some authors [1 -5] have stated that, although the diaphyses and meta-physes can be affected in both conditions, periostitis commonly extends into theepiphyseal region in primary HOA. Poorly defined bony outgrowths are also saidto be characteristic of primary HOA, differing from the linear deposits that mosttypically accompany secondary HOA [1 -5]. Other investigators [6-1 0] have foundno differences in the radiographic appearance of periostitis in the two types ofHOA.

The purpose of this investigation is to characterize the thickness, morphology,and distribution of periostitis in primary and secondary HOA and to identify featuresthat differentiate the disorders. The study will also test the hypothesis that thepattern of periostitis in HOA is dependent not on its primary or secondary naturebut principally on the duration of the disease process.

Subjects and Methods

Our study population consisted of 24 patients who had one of the following diseases: (1)primary or idiopathic hypertrophic osteoarthropathy (primary HOA), (2) cyanotic congenitalheart disease with secondary HOA (CCHD-associated HOA), or (3) lung cancer associated

774 PINEDA ET AL. AJA:148, April 1987

Note-Numbers in parentheses are percentages. HOA = hypertrophic osteoarthropathy; CCHD = cyanotic congenital heart disease.

with secondary HOA (lung-cancer HOA). The primary HOA patients

included six men ranging in age from 8 to 42 years (mean ± SD =

26 ± 1 2 years), with disease duration ranging from 3 to 28 years(mean ± SD = 1 1.2 ± 5.8 years). The CCHD-associated HOA groupcomprised 10 men and one woman ranging in age from 10 to 52years (mean ± SD = 25.6 ± 1 0.8 years), with disease durationranging from 5 to 52 years (mean ± SD = 21 .7 ± 1 0.9 years).Underlying congenital cardiac defects included transposition of thegreat vessels in four patients, patent ductus arteriosus with flowreversal in two, anomalous pulmonary venous connection in two,

atrioventricular discordance with situs inversus in one, a single yen-tricle in one, and tetralogy of Fallot in one. The lung-cancer HOApatients included seven men ranging in age from 48 to 67 years(mean ± SD = 57.7 ± 6.3 years), with disease duration ranging from1 to 7 years (mean ± SD = 2.8 ± 2.1 years). Histologic subtypes

included adenocarcinoma in four patients, epidermoid carcinoma inone, bronchoalveolar carcinoma in one, and squamous cell carcinomain one.

The predominance of males (23 of 24) in the study reflected thenatural tendency of primary HOA and lung cancer to affect thisgender. There appeared to be no major difference in age betweenthe groups of patients with primary HOA and CCHD-associated HOA;nevertheless, the average duration of disease was greater in thepatients with CCHD-associated HOA. The seven patients with lung-cancer HOA were the oldest, and the duration of their disease wasthe shortest.

Our criteria for diagnosing HOA required the combined presence

ofdigitalclubbing and periosteal new-boneformation in tubular bones.Primary HOA was diagnosed when after extensive clinical and labo-ratory analysis the aforementioned abnormalities were evident in theabsence of any associated organic disease.

Skeletal surveys were obtained in all patients with primary HOAand CCHD-associated HOA and included frontal radiographs of thelong bones of the upper and lower extremities as well as those of thehands and feet. Skeletal surveys from patients with lung-cancer HOAwere selected from the teaching files of the radiology department; in

all of these surveys, radiographs of the upper or lower extremities orboth were available.

The radiographs were analyzed and graded by two experienced

skeletal radiologists without knowledge of the clinical diagnosis. Thetwo radiologists first developed a set of specific observations to bemade in interpreting the radiographs, which are outlined below. Asubset of the films were then graded independently by both observ-ers, and no significant interobserver variation was encountered. Oneof the two observers then graded all of the remaining radiographs

included in the study.

The radiographic analysis of each long and short tubular boneincluded the following observations: (1) site of involvement: presence

or absence of periosteal new-bone formation in the diaphysis, meta-physis, or epiphysis; and (2) pattem of periostitis: (a) single layer ofnew-bone formation with a radiolucent area between this bone andsubjacent cortex, (b) multilayered apposition of several layers of new

bone with either a radiolucent area between each layer or a thick,linear, osseous deposit that was separated from or merged with theunderlying bone, (c) irregular pattern, characterized by thick, irregular,and diffuse osseous proliferation that obscures the original corticalsurface and leads to a pseudocortex with calcification of the neigh-boring structures including tendons, ligaments, or interosseous mem-branes. On the basis of the observations outlined above, the overall

severity of periostitis was subjectively graded as mild (1 +), moderate(2+), or severe (3+).

In order to further evaluate the degree of periostitis, accurate

measurement of the thickness of the cortical bone and medullaryspace in the middiaphyseal region of each second metacarpal wasperformed by using standard radiogrammetric technique. Statisticalanalysis ofthis quantitative data was performed by using a completelyrandomized least-squares analysis of variance.

Results

Radiographic features related to the site of involvement,type of periostitis, and overall subjective grade of severity aresummarized in Table 1 . All three segments of the bone(diaphysis, metaphysis, and epiphysis) displayed periostealnew-bone formation in the group of patients with primaryHOA and CCHD-associated HOA (Fig. 1); none of the patientswith HOA secondary to lung cancer showed epiphyseal ab-normalities.

The most common type of periostitis observed in patientswith primary HOA (Fig. 2) and CCHD-associated HOA (Fig. 3)was multilayered; an irregular pattern of periostitis was pres-ent in one-third of these patients (Figs. 4 and 5). In the patientswith lung-cancer HOA, however, the most common patternof periostitis was a single layer (Figs. 6 and 7), and in no casewas an irregular type of periostitis found.

Scores indicating severity of periostitis showed that a se-vere degree was present in a significant percentage of pa-tients with both primary HOA and CCHD-associated HOA.None of the patients with HOA secondary to lung cancerexhibited such a score.

TABLE 1: Radiographic Abnormalities of Tubular Bones in Patients with Hypertrophic Osteoarthropathy

Primary HOA(n=6)

CCHD-Associated HOA(n=11)

Lung-Cancer HOA(n=7)

Type of periostitis

Single layer 1 (17) 2 (1 8) 4 (57)Multilayered 3 (50) 6 (55) 3 (43)Irregular 2 (33) 3 (27) 0

Site of involvementDiaphysis 2 (33) 4 (36) 3 (43)Diaphysis, metaphysis 2 (33) 3 (27) 4 (57)Diaphysis, metaphysis, epiphysis 2 (33) 4 (36) 0

Overall severity indexGrade 1 + (mild) 2 (33) 5 (45) 3 (43)Grade 2+ (moderate) 2 (33) 4 (36) 4 (57)Grade 3+ (severe) 2 (33) 2 (18) 0

AJA:148, April 1987 PERIOSTITIS IN HYPERTROPHIC OSTEOARTHROPATHY 775

Fig. 1.-Frontal radiograph of right hand of 24-year-old man with anomalous pulmonary venousconnection shows penosteal new-bone formationinvolving metacarpals, proximal and middle pha-langes, and radial styloid process. Periostitis isfirmly incorporated into underlying cortex and cx-tends into epiphyses at two sites (arrows), typicalof longstanding disease.

Fig. 2.-Frontal radiograph from patientwith primary hypertrophic osteoarthropathydepicts apposition of several layers of newbone (arrows) along cortical surface of distalhumerus, characteristic of early age of onsetin this individual.

- � I-

I�. �T�’�-e�r1 � �:::� ___-�_-

Fig. 3.-Frontal radiograph of distal tibia andfibula from 20-year-old man with situs inversusand athoventricular discordancy shows multilay-ered, thick, andirregularosseous proliferation withspiculated contour (arrow) that merges with un-denying cortex. Pattern resembles that tradition-ally associated with primary hypertrophic osteo-arthropathy.

Fig. 4.-Frontal radiograph from 34-year-old patient with primary hypertrophic osteo-arthropathy shows characteristic features oflongstanding disease. Severe periostitis in-volves distal tibia and fibula, manifested asthick, irregular osseous proliferation (ar-rows) that obscures original cortical surface.Adjacent structures including interosseousmembranes (n) and entheses are also af-fected.

Fig. 5.-Frontal radiograph of radial andulnar shafts from 25-year-old patient withcyanotic heart disease shows nodular andirregular appearance of periosteal new-boneformation (arrows), a pattern not observed inassociation with bronchogenic carcinoma.

Fig. 6.-Frontal radiographs of distal portions of A, right femur and B, left tibia from 66- Fig. 7.-Frontal radiograph of distal radius and ulna fromyear-old man with lung carcinoma show solitary layers of periosteal new-bone formation 52-year-old man with epidermoid carcinoma of right upper(arrows) with interposed radiolucent areas separating them from subjacent cortex. lobe shows single thick layers of periosteal new bone (ar-

rows) separated from underlying cortices by radiolucentstripes. This periosteal morphology occurs with relativelylow frequency in primary hypertrophic osteoarthropathy(HOA) and HOA associated with cyanotic congenital heartdisease.

776 PINEDA ET AL. AJA:148,April 1987

TABLE 2: Disease Duration in Years (Mean ± SD) Compared to Overall Severity of Radiographic Abnormalities in Patients withHypertrophic Osteoarthropathy

Severity of RadiographicAbnormalities

Primary HOA(n = 6)

CCHD-Associated HOA(n= 1 1)

Lung-Cancer HOA(n = 7)

Gradel+(mild) 10 ± 2.8 17.2± 9.8 3.1 ±2.5Grade2+(moderate) 13.5±14.8 21.5± 2.1 3.5 ±2.0Grade3+(severe) 16.5± 2.1 31.6±13.5 0

Note.-HOA = hypertrophic osteoarthro pathy; CCHD = cyanotic congeni tal heart disease.

There was a direct relationship between the duration ofdisease and the morphology as well as the extent of theradiographic abnormalities (Table 2), and between the diseaseduration and the number of affected bones (Table 3) in thosepatients with primary HOA and CCHD-associated HOA forwhom complete skeletal surveys were available. When theradiographs from patients in all three groups were matchedfor disease duration, a similar pattern of periostitis and asimilar index of severity were observed.

Table 4 shows the frequency with which each bone was

involved. The tibia and fibula were most commonly affected,followed by the femur and ulna. With the exception of threepatients with primary HOA, the abnormalities were bilateraland symmetric in distribution.

Statistical analysis of metacarpal measurements revealed

significant differences between the patients with lung-cancerHOA and the other two groups (Table 5). The former mani-fested lower values for cortical-bone (p < .0001) and total-bone (p < .0001) width as well as higher values for medullary-space width (p < .0014).

AJA:148, April 1987 PERIOSTITIS IN HYPERTROPHIC OSTEOARTHROPATHY 777

Discussion

The results of this study indicate that the skeletal changesobserved in primary HOA and in CCHD-associated HOA aresimilar and are characterized by more prominent and wide-

spread abnormalities than those seen in lung-cancer HOA.The radiographic differences related to thickness, morphol-ogy, and distribution of periostitis that were observed amongthese groups appeared primarily to reflect the duration of thedisease. For example, young patients with either primary HOAor CCHD-associated HOA had a pattern of periostitis similarto that in adults with lung-cancer HOA. Unfortunately, thenature of our available study material did not allow us to

ascertain whether or not these generalizations are applicableto patients with other causes of HOA.

Some investigators [1 , 2, 5, 1 1] have indicated that a carefulevaluation of radiographic abnormalities will allow accurate

TABLE 3: Relationship of Number of Affected Bones to DiseaseDuration

Disease Duration(years)

Primary HOA CCHD-Associated HOA

1-10 5 911-20 21 1121-30 52 21

31-40 0 041-52 0 31

Note.-HOA = hypertrophic osteoarthropathy; CCHD = cyanotic congenital

heart disease.

TABLE 4: Distribution of Affected Bones in Patients with PrimaryHypertrophic Osteoarthropathy and Cyanotic Congenital HeartDisease with Secondary Hypertrophic Osteoarthropathy

Affected BoneNumber of patients

( /0)

Tibia 14(82)Fibula 13(76)Femur 12(71)

lAna 12(71)Metatarsal 1 1 (65)Radius 9 (53)Metacarpal 9 (53)Humerus 8(47)Hand phalanges (proximal and middle) 5 (29)Foot phalanges (proximal and middle) 4(24)

Note-n = 17.

differentiation between primary and secondary HOA. For ex-ample, it has been suggested that in secondary HOA, theepiphyses are not affected by periosteal new-bone formation[1 , 2, 5, 1 1 ]. Our results have shown that this segment of thetubular bones is affected in one-third of patients with CCHD-associated HOA and in a similar proportion of patients withprimary HOA. It has also been stated that another basicdifference is the type of periostitis, being “layered” in second-ary HOA and irregular in primary HOA [3, 5]. Our results donot support this concept; we found shaggy, irregular, perios-teal bone formation equally frequent in both primary HOA andCCHD-associated HOA (Table 1). This type of periosteal newbone was not detected in the lung-cancer HOA group. A thirddistinguishing feature between the two types of HOA notedby some investigators is the presence of ossification in liga-ments or interosseous membranes among patients with pri-mary HOA [5]. We found three examples of such ossificationbetween the tibia and fibula; two occurred in patients withCCHD-associated HOA and the other in a patient with primaryHOA. These three patients exhibited widespread bony in-volvement and had the longest duration of disease.

Our investigation did not disclose any distinguishing radio-graphicfeatures between primary HOA and CCHD-associatedHOA. A similar percentage of patients in both groups dis-played the same type of periosteal new-bone formation. Bothconditions affected all three segments of tubular bones withsimilar frequency, and the thickness as well as the morphol-ogy of the periostitis was similar, supporting the observationsof yogI and Goldfischer [1 2], who compared the histologicfeatures in cases of primary and secondary HOA [13-15].They found strikingly similar changes in both conditions,including identical patterns of periosteal new-bone formation,the same gradual conversion of newly formed cancellous boneinto a compact outer sheath, and comparable progressiverarefaction of the compact outer layers of the original cortex.

In our seven patients with lung-cancer HOA, less prominentperiosteal changes were noted in a more limited distribution.When radiographs were analyzed according to disease du-ration, we found no significant differences between this groupof patients and those with primary and CCHD-associatedHOA; these observations suggest that the duration of thedisease is most important in determining the distribution andmorphology ofthe skeletal changes. Therefore, in malignancy-associated HOA, a disease process of much shorter duration,the natural evolution of the osseous changes cannot fully beexpressed.

TABLE 5: Radiogrammetric Measurements in mm (Mean ± SD) of Second Metacarpal in Patients wHOA, and Lung-Cancer HOA

ith Primary HOA, CCHD-Associated

Primary HOA CCHD-Associated HOA Lung-Cancer HOA(n=6) (n=11) (n=7)

Cortical-bone width 8.9 ± 6.0 8.5 ± 6.4 6.0 ± 3.9Medullary-space width 1 .2 ± 5.3 1 .7 ± 5.7 3.0 ± 3.8Total-bone width 9.7 ± 1 .0 9.8 ± 1 .7 8.9 ± 2.4

Note.-HOA = hypertrophic osteoarthropathy; CCHD = cyanotic congenital heart disease.

778 PINEDA ET AL. AJR:148, April 1987

In most of our patients, regardless of the cause of HOA,the skeletal abnormalities were symmetric in distribution. Thebones of the lower extremities were most frequently affected,a finding that is in agreement with previous observations [1 1,16-20].

Our data suggest that in all forms of HOA, periosteal bonedeposition appears initially in the proximal and distal dia-physes and metaphyses of the long tubular bones. It consistsof monolayered, linear periostitis that increases the circum-ference of the affected bone without otherwise altering itsshape. With disease progression, periostitis becomes moreprominent, extends into the epiphyses, and is laminated ormultilayered in appearance. In advanced cases, the changescan affect virtually all of the long tubular bones, resulting inan increased thickness and an irregular surface. This processis reflected by the group differences in metacarpal corticaland medullary thicknesses derived from the quantitative por-tion of this investigation. Ossification of an adjacent interos-seous membrane appears to be the final step of the evolutionof HOA.

One potential limitation of this investigation involves thefact that age of onset and duration of disease were notindependent variables among the patients studied. Long du-ration of disease was associated with early age of onset(primary HOA and CCHD-associated HOA), whereas shortduration of disease tended to occur in the older patients withHOA secondary to bronchogenic carcinoma. Our results donot permit exclusion of the possibility that age of onset asopposed to duration of disease is the predominant factor indetermining the pattern and distribution of HOA.

The loosely attached periosteum of children may be moreconducive to new-bone formation, resulting in a greater ten-dency toward thick, shaggy periostitis with epiphyseal exten-sion. This hypothesis could be addressed by a comparableradiographic study of a population with HOA secondary to anadult-onset disease that is compatible with a longer lifespan,such as bronchiectasis. Unfortunately, skeletal surveys on asufficient number of such patients were not available at thetime of our investigation.

In conclusion, the periostitis of primary HOA and thatsecondary to CCHD are similar in appearance and character-ized by thicker, more irregular, and widespread changesinvolving all of the segments of a tubular bone. By contrast,lung-cancer HOA typically includes a mild degree of periostitisthat spares the epiphyses and is monolayered. Correlation ofradiographic patterns with duration of disease confirms thatmore prominent and widespread changes are indicative oflongstanding disease. The periosteal manifestations of HOAthus appear to be principally dependent not on the primary or

secondary nature of the disease but rather on its duration orage of patient at onset.

ACKNOWLEDGMENTS

Special thanks are extended to Carol Latham and Cynde Rochefor outstanding secretarial assistance.

REFERENCES

1 . Bhate Dv, Pizarro AJ, Greenfield GB. Idiopathic hypertrophic osteoarthrop-athy without pachyderma. Radiology 1978;129:379-381

2. Greenfield GB. Radiology of bone diseases, 2nd ed. Philadelphia: Uppin-

cott, 1975:383-3893. Lazarus HJ, Galloway JK. Pachydermoperiostosis: an unusual cause of

finger clubbing. AJR 1973;1 18:308-3134. Brugsch HG. Acropachyderma with pachypenostosis: report of a case.

Arch Intern Med 1941;68:687-7005. Aesnick D, Niwayama G. Diagnosis of bone and joint disorders with

emphasis on articular abnormalities. Philadelphia: Saunders, 1981:2983-2996

6. Hedayati H, Barmada A, Skosey JL. Acrolysis in pachydermoperiostosis:primary or idiopathic hypertrophic osteoarthropathy. Arch Intern Med1980:140(8): 1087-1 088

7. Harbison JB, Nice CM Jr. Familiar pachydermopenostosis presenting as

an acromegaly-like syndrome. AIR 1971;1 12:532-5368. Metz EN, Dowell A. Bone marrow fallure in hypertrophic osteoarthropathy.

Arch Intern Med 1965;1 16:759-7649. Keats TE, Bagnall WS. Chronic idiopathic osteoarthropathy. Radiology

1954;62:841-84410. Hecht A. Idiopathic hypertrophic osteoarthropathy. NY State J Med

1965;65:3038-30441 1 . Greenfield GB, Schorsch HA, Shkolnik A. The various roentgen appear-

ances of pulmonary hypertrophic osteoarthropathy. AiR 1967;1O1:927-931

12. yogI A, GOIdfiSCher S. Pachydermopenostosis. Primary or idiopathic hy-

pertrophic osteoarthropathy. Am J Med 1962;33:166-187

13. Uehlinger E. Hyperostosis generalisata mit Pachydermi. Virchows Arch(Pathol Anat) 1942;308:396-444

14. Crump C. Histologie der Allgemeinen Osteohytose (OsteoarthropathieHypertrophiante Pneumique). Virchows Arch (Pathol Anat) 1929;271:

467-51115. Gall E, Bennett GA, Bauer W. Generalized hypertrophic osteoarthropathy:

a pathologic study of seven cases. Am J Pathol 1951;27:349-38116. Pineda CJ, Guerra J Jr, Weisman MH, Aesnick D, Martinez-Lavin M. The

skeletal manifestations of ciubbing: a study in patients with cyanoticcongenital heart disease and hypertrophic osteoarthropathy. Semin Arthri-tis Rheum 1985;14:263-273

17. Martinez-Lavin M, Bobadila M, Casanova J, Attie F, Martinez M. Hyper-

trophic osteoarthropathy in cyanotic congenital heart disease: its preva-

lence and relationship to bypass of the lung. Arthritis Rheum1982;25: 1186-1193

18. Martinez-Lavin M, Amigo MC, Castillejos G, Padilla L, vintimilla F. Coex-istent gout and hypertrophic osteoarthropathy in patients with cyanoticheart disease. J Rheumatol 1984;1 1(6):832-834

19. All A, Tetalman MA, Fordham EW, et al. Distribution of hypertrophicpulmonary osteoarthropathy. AiR 1980;134:771-780

20. Aesnick D. Radiographic features of osteoarthropathy (letter). Radiology1985;1 57:553-554