Volume 23Number 6, Part 1December 1990

pityriasis rubra pilaris (PRP) with seborrheic keratoses.The authors review the clinical and histologic features ofPRP and the previous and current systemic therapeuticmodalities. Moreover, they reported Griffiths' classifica­tion of PRP (based on age of onset, cutaneous features,disease course and prognosis); classic adult (type I),atypical adult (type II), classic juvenile (type III), cir­cumscri~ed juvenile (type IV), and atypical juvenile(type V).

Our experience differs from this classification. In ourstudy of 29 cases of PRP in childhood,1 two principalfindings have been observed. First, in no patient was therea family historyofPRP; because PRP has been describedas an autosomal dominant disease,2-4 it is surprising thatwe have not observed (in 20 years) any cases ofhereditaryPRP. We believe, in agreement with Griffiths' con­clusions,5that many cases of hereditary PRP are insteadatypical forms ofichthyosis or psoriasis. Second, we couldnot establish a correlation between clinical features andprognosis in juvenile PRP (as proposed by Griffiths' clas­sification). In fact, according to Griffiths' criteria, 10 ofour 29 patients could be included in the classic juvenileform. However, in none did onset occur during the first orsecond year of life (mean age 4.4 years) nor were thereerythrodermic forms. Moreover, the dermatosis healedwithin 3 years in 90% of cases (compared with 16% inGriffiths' group). Again with Griffiths' criteria, 19 ofourpatients could be included in the circumscribed juvenileform of PRP. However, their age of onset was not theprepubertal age (mean age 6.3 years) and the dermatosishealed within 3years in 68% of cases (D. Cerri, PhD the­sis, unpublished data) (compared with 32% of Griffiths'patients). Thus it may not be useful to classify juvenilePRP by age of onset and cutaneous features used asprognostic factors. We proposed a classification ofchild­hood PRP in three forms: acute (resolves in less than 6months), acute with a prolonged course (resolves within1year), and chronic (persists more than 1year).l All threeforms of juvenile PRP can have localized or diffuselesions. In short, on the basis of our findings, the extentand the location of the lesions and the age ofonset do nothelp to establish a prognosis for PRP, at least in children.

Carlo Gelmetti, MD, and Danilo Cerri, MD,I Dermatologic Department and Pediatric

Dermatology, University ofMilan, Milan, Italy

REFERENCES

1. Gelmetti C, Schiuma AA, Cerri D, et al. Pityriasis rubra pi·laris in childhood: a long-term study of 29 cases, PediatrDermatoI1986;3:446-51.

2. Goldsmith LA, Weinrich AE, Shupack MD. Pityriasis rubrapHaris response to 13-cis-retinoic acid (isotretinoin). ] AMACAD DERMATOL 1982;6:710-5.

3. Kint A, De Bie S, Geerds ML, et al. Pityriasis rubra pHaris,a familial condition. Arch Belg DermatoI1972;158:371-6.

Correspondence 1187

4. Touraine A. L'eredite dans Ie pityriasis rubra pilaire. AnnDermatol VenereoI1942;8:175-7.

5. Griffiths WAD. Pityriasis rubra pilaris. Clin Exp Dermatol1980;5:105-12.

Reply

To the Editor: We are pleased that our article on pityr­iasis rubra pilaris (PRP) prompted Drs. Gelmetti andCerri to discuss their experience regarding PRP in child­hood; we appreciate their constructive criticisms of thecurrent classification (Griffiths') for this disease and theirproposed classification ofchildhood PRP.l Griffiths' clin­ical classification of PRP was originally published as anabstract in 1977; at that time, five categories were dermedaccording to the clinical features and prognosis observedin a studyof93 patients with PRP.2 In 1980, on the basisof the findings in more than 100 of his own PRP patientsand a critical evaluation of the previously publishedliterature,3,4 Griffiths elaborated on these five groups ofpatients and the ability to classify PRP patients on thebasis of clinical appearance, behavior, age of onset, prog­nosis, and possibly etiology.5 Briefly, the majority of pa­tients were adults with classical (type I) features. Chil­dren comprised 45% of the PRP patients: the circum­scribed (type IV) juvenile form was the most commonlyobserved type and occurred 2.5 times as often as the clas­sical (type III) juvenile form. The atypical juvenile form(type V), similar to the atypical adult form (type II), wasinfrequently observed; furthermore, presumptively basedon the uncharacteristic features, Griffiths also postulatedthat these atypical forms (types II and V) of PRP mightreally represent an "ichthyosis-like disorder" or an "atyp­ical follicular ichthyosis."5 Although approximately 30families with "hereditary PRP" have been reported since1975, the majority ofPRP cases are sporadic.3,4 In fact,a family history of PRP is uncommon and has usuallyonly been described in patients with the atypical juvenileform of PRP.2.5

The retrospective study of PRP in childhood byGelmetti et al. included long-termfollow-up evaluation of29 cases; in addition to not observing any cases of hered­itary PRP, they were also unable to establish a relationbetween the severity ofPRP and its prognosis.1 Whetherclassified according to Griffiths' criteria or considered asa single group, the percentage of children in whom thedermatosis resolved within 3 years of disease onset wassignificantly higher in the group studied by Gelmetti etal. l compared with that evaluated by Griffiths5; the treat­ment modalities used for each individual patient were notdescribed in either study. After analyzing their cases,Gelmetti et al. proposed several primary and secondarydiagnostic criteria and a classification system (basedsolely on the duration of active disease) for childhoodPRP. Unfortunately, although Gelmettietal. criticize thelack of correlation between clinical features and progno-

1188 Correspondence

sis with the use of Griffiths' classification of childhoodPRP for their patients, their proposed classificationsystem does not enable the clinician to establish a prog­nosis after diagnosing new-onset PRP in a child.

Although we agree that the clinical features and prog­nosis in the children with PRP described by Gelmetti etal. did not show a correlation when the patients weregrouped according to Griffiths' classification, we questionthe prognostic usefulness of their classification system.Griffiths' original classification of childhood PRP mightbe modified to reflect the more favorable experience ofGelmetti et al. in the resolution of childhood PRP. Addi­tional prospective evaluation of the onset age, lesion mor­phologyand distribution, response to therapy, and diseaseduration in adults and children with PRP may clarify therelation (if present) between clinical features and diseaseprognosis in this condition.

Philip R. Cohen, MD,a and Janet H. Prystowsky,MD, PhD,b Pasadena Dermatology Clinic, Pasadena,Texas,a and Department ofDermatology, College of

Physicians and Surgeons of Columbia University,New York, New Yor0

REFERENCES

1. Gelmetti C, Schiuma AA, Cerri D, et al. Pityriasis rubra pi­laris in childhood: a long-term study of 29 cases. PediatrDermatol 1986;3:446-51.

2. Griffiths WAD. Pityriasis rubra pHaris: clinical features andnatural history in a study of 93 patients. Br J Dermatol1977;97(suppl 15):18.

3. Griffiths WAD. Pityriasis rubra pilaris-an historical ap­proach. Trans St Johns Hasp Dermatol Soc 1975;61 :58-69.

4. Griffiths WAD. Pityriasis rubra pilaris-an historical ap­proach. II. Clinical features. Coo Exp DermatolI976;1:37­50.

5. Griffiths WAD. Pityriasis rubra pilaris. Clin Exp Dermatol1980;5:105-12.

Similar high frequency of IgA antireticulin andantiendomysial antibodies in dermatitisherpetiformis

To the Editor: We read with interest the recent article byPeters and McEvoy on the occurrence of IgA-classantiendomysial antibodies (EMA) in patients with der­matitis herpetiforrnis (DR) (J AM ACAD DERMATOL1989;21:1225-31). They examined serum from 24 pa­tients withDH. Ofthe21 patients receiving a normaldiet,19 (90%) had positive circulating IgA-EMA in titers thatranged from 1:5 to 1:320. Serum from three patients withDR after gluten-freediets and 77 of80 (96%) controlserafrom patients with cutaneous and noncutaneous diseaseshowed negative results. The sensitivity of IgA-EMA testwas 90% and the specificity 96% when DR patients notfollowing gluten-free diets were considered.

Unfortunately, no jejunal biopsies were performed in

Journal of theAmerican Academy of

Dermatology

the study, but in previous studies the occurrence of IgA­EMA in patients with DR have been shown to correlatewith the presence ofjejunal villous atrophy. 1, 2 Thereforethe two IgA-EMA-negative DH patients with normaldiets reported byPeters and McEvoy mighthave had onlyslight or no small-intestinal abnormalities. In agreementwith this suggestion, we have found in our studies nega­tive IgA-EMA in 3 of 11 (27%) DH patients with partialvillous atrophy and in 8 of 13 (62%) patients with DHshowing slight changes or normal jejunal mucosa.I, 3 Theoverall frequencies ofnegative 19A-EMA in DR patientswith normal diets have been 24% and 17%, respectively,in our previous studies. Because of several negativeresults, the use of the IgA-EMA test cannot be recom­mended as an diagnostic aid in DB, but we agree that thistest is helpful when examining DR patients with jejunalvillous atrophy and after their treatment with a gluten­free diet.

Peters and McEvoy mentioned in their article that IgAantireticulin antibodies (ARA) occur in a variable butgenerally small proportion of patients with DH. In con-

. trast to this statement, we had reported in 1985 that 19A­ARA test is a good detector ofjejunal damage in patientswith DH.4 Of 25 DR patients with subtotal villous atro­phy 24 (96%) had these antibodies. Moreover, the IgA­ARA test became negative in every patient treated witha gluten-free diet. We havepreviously also shown thehighsensitivity and specificity of IgA-ARA test in the detec­tion ofceliacdisease in children,5 and recently we showedthat the same is true for adult celiac disease.3 In the latterstudy, none of the control subjects with ulcerative colitisor Crohn's disease had IgA-ARA and the only IgA­ARA positive subject among 100 blood donors had undi­agnosed celiac disease. When we compared the occur­rence'ofIgA-ARA and IgA-EMA in DR and celiac dis­ease, the frequencies were almost identical; the titers werealso comparable to each other and declined in parallelduring gluten-free diet treatment.3 IgA-ARA and IgA·EMA tests also gave comparable results when asymp­tomaticceliac disease was searched in relatives ofpatientswith celiac disease.6 Moreover, the absorption studies werecently performed detected a human subtype of ARAsimilar to EMA, but more detailed experiments areneeded to confirm the identity of these antibodies.3

Presently the IgA-EMA and IgA-ARAtests are basedon immunofluorescence techniques. The former usesmonkey esophagus and the latter rat kidney and liversections as substrate. When developing our IgA-ARAtest, we found it important to use a good polyvalent an­tihuman immunoglobulin conjugate or to use anti-IgAconjugate in primary screening of the serum samples.ARAcan be exclusively of19A class, and it seems that inprevious studies thebelief that ARA is mostly ofIgGclasshas guided the selection ofconjugates and resulted in lowfrequency of positivity. When the IgA-EMA test is used,it is important to have tissue from the lower third of the