Formative evaluation of ontology learning methods for entity discovery by using existing ontologies as reference standards.

OBJECTIVE: Developing a two-step method for formative evaluation of statistical Ontology Learning (OL) algorithms that leverages existing biomedical ontologies as reference standards. METHODS: In the first step optimum parameters are established. A 'gap list' of entities is generated by finding the set of entities present in a later version of the ontology that are not present in an earlier version of the ontology. A named entity recognition system is used to identify entities in a corpus of biomedical documents that are present in the 'gap list', generating a reference standard. The output of the algorithm (new entity candidates), produced by statistical methods, is subsequently compared against this reference standard. An OL method that performs perfectly will be able to learn all of the terms in this reference standard. Using evaluation metrics and precision-recall curves for different thresholds and parameters, we compute the optimum parameters for each method. In the second step, human judges with expertise in ontology development evaluate each candidate suggested by the algorithm configured with the optimum parameters previously established. These judgments are used to compute two performance metrics developed from our previous work: Entity Suggestion Rate (ESR) and Entity Acceptance Rate (EAR). RESULTS: Using this method, we evaluated two statistical OL methods for OL in two medical domains. For the pathology domain, we obtained 49% ESR, 28% EAR with the Lin method and 52% ESR, 39% EAR with the Church method. For the radiology domain, we obtain 87% ESA, 9% EAR using Lin method and 96% ESR, 16% EAR using Church method. CONCLUSION: This method is sufficiently general and flexible enough to permit comparison of any OL method for a specific corpus and ontology of interest.

Effectiveness of lexico-syntactic pattern matching for ontology enrichment with clinical documents.

OBJECTIVE: To evaluate the effectiveness of a lexico-syntactic pattern (LSP) matching method for ontology enrichment using clinical documents. METHODS: Two domains were separately studied using the same methodology. We used radiology documents to enrich RadLex and pathology documents to enrich National Cancer Institute Thesaurus (NCIT). Several known LSPs were used for semantic knowledge extraction. We first retrieved all sentences that contained LSPs across two large clinical repositories, and examined the frequency of the LSPs. From this set, we randomly sampled LSP instances which were examined by human judges. We used a two-step method to determine the utility of these patterns for enrichment. In the first step, domain experts annotated medically meaningful terms (MMTs) from each sentence within the LSP. In the second step, RadLex and NCIT curators evaluated how many of these MMTs could be added to the resource. To quantify the utility of this LSP method, we defined two evaluation metrics: suggestion rate (SR) and acceptance rate (AR). We used these measures to estimate the yield of concepts and relationships, for each of the two domains. RESULTS: For NCIT, the concept SR was 24%, and the relationship SR was 65%. The concept AR was 21%, and the relationship AR was 14%. For RadLex, the concept SR was 37%, and the relationship SR was 55%. The concept AR was 11%, and the relationship AR was 44%. CONCLUSION: The LSP matching method is an effective method for concept and concept relationship discovery in biomedical domains.

Lymphomatoid papulosis type A: clinical, morphologic, and immunophenotypic study.

BACKGROUND: Lymphomatoid papulosis (LyP) is a cutaneous clonal or polyclonal Ki-1 + T-cell lymphoproliferative disorder, morphologically resembling Ki-1 + anaplastic large cell lymphomas (Ki-1 + ALCL) or Hodgkin's disease (HD). Lymphomatoid papulosis usually has a characteristic benign clinical course with remissions and relapses of the cutaneous eruptions. METHODS: The authors studied three patients with LyP. In each case the diagnosis was established based on the typical clinical history and presentation of the cutaneous lesions as well as the morphologic and immunophenotypic findings. RESULTS: In all three cases the skin biopsies showed a polymorphic, nonepidermotropic, dermal lymphocytic infiltrate, composed of small lymphocytes and fewer large, atypical cells. The large cells were positive for the activation markers CD30 (Ki-1) and CD45R (leukocyte common antigen), and were negative for the HD marker CD15 (Leu MI). CONCLUSIONS: In most cases, LyP can be distinguished from Ki-1 + ALCL and HD on the basis of clinical, morphologic, and/or immunophenotypic findings. We emphasize the importance of the recognition of LyP as a clinicopathologic entity and the awareness of dermatologists, oncologists, and surgical pathologists in differentiating LyP from other primary cutaneous Ki-1 + lymphoproliferative disorders (Ki-1 + ALCL and HD). The prognosis of cutaneous Ki-1 + ALCL and HD is usually different from LyP and requires a different therapeutic approach.

T-cell-rich B-cell lymphoma presenting in skin. A clinicopathologic analysis of six cases.

We reviewed our experience with six T-cell-rich B-cell lymphomas (TRBL) presenting in skin. Immunohistochemical studies were performed on all biopsies. The lymphoid population consisted mainly of CD3 and/or UCHL-1 (CD45RO) positive T cells. 5 to 15% of the lymphoid cells stained for the B-cell marker L26 (CD20). Monoclonality of the B-cell component was demonstrated in all cases, utilizing either light chain restriction (5 cases) or clonal immunoglobulin heavy chain gene rearrangement by polymerase chain reaction (PCR) (2 cases). One case was confirmed to be monoclonal by both techniques. Additionally, no clonal rearrangements of the T-cell receptor gamma gene were observed. There was considerable morphological variety in these cases. In H&E stained sections, the differential diagnosis included pseudolymphoma, peripheral T-cell lymphoma, Hodgkin's disease, Lennert's lymphoma and a MALT lymphoma. A significant component of monoclonal plasma cells was present in 3 of 6 cases, suggesting a possible origin from cutaneous immunocytoma. In fact, one of our cases was a biphasic lymphoma displaying TRBL with a small focus of immunocytoma. We conclude that immunophenotypic analysis is necessary for the diagnosis of TRBL. Pathologists should be aware of this type of cutaneous B-cell lymphoma to avoid misinterpretation as a pseudolymphoma.

Polymerase chain reaction versus Southern blot hybridization. Detection of immunoglobulin heavy-chain gene rearrangements.

To determine efficiently the clonality of B-cell lymphoproliferative disorders, we modified an immunoglobulin heavy-chain (IGH) gene rearrangement polymerase chain reaction (PCR) assay that requires only a single primer germline variable (VH) and joining (JH) pair and does not involve nested priming, blot hybridization, radioactivity, or sequencing of the amplified PCR product. This simple PCR technique enabled detection of IGH gene rearrangements in as little as 10 pg (one cell equivalent) of DNA or when the clonal-to-polyclonal B-cell ratio was experimentally set at 1:1000. We detected IGH gene rearrangements in 83.5% (71 of 85) of clonal B-cell processes, a sensitivity approaching that of more cumbersome multiple primer and nested primer assays. Moreover, this technique is equally effective with fixed tissues, either B5 or formalin, and can be performed on minute samples, histologic sections, fine-needle aspirates, or cerebrospinal fluids. When compared with conventional Southern blot analysis using a genomic JH probe, the PCR assay demonstrated IGH gene rearrangements in 82% (37 of 45) of B-cell processes positive by Southern blot. No false-positive results were observed in 29 negative control tissues. We now use IGH gene PCR routinely in our laboratory for the detection of clonal B-cells in virtually any tissue sample as an aid in early diagnosis, staging, and monitoring, and the Southern blot procedure is reserved for only a minority of diagnostic cases. for only a minority of diagnostic cases.

Constitutive secretion of soluble interleukin-2 receptor by human T cell lymphoma xenografted into SCID mice. Correlation of tumor volume with concentration of tumor-derived soluble interleukin-2 receptor in body fluids of the host mice.

Increased serum concentration of soluble alpha-chain receptor for interleukin-2 (sIL-2R) has been noted in patients with a variety of inflammatory conditions and lymphoid malignancies including T cell leukemia and lymphoma. Elevated sIL-2R serum levels seen in lymphoid malignancies appear to correlate with the clinical stage of disease. However, because sIL-2R is produced by normal activated lymphocytes, it has been uncertain whether serum sIL-2R in such conditions is derived from tumor cells or normal immune cells responding to the tumor. To address this question, we used a model of human (CD30+) anaplastic, large T cell lymphoma transplanted into immunodeficient SCID mice. Reverse transcription polymerase chain reaction of tumor RNA showed that the tumor, designated mJB6, contains mRNA for alpha-chain of human IL-2R. Furthermore, 15 to 25% of tumor cells stained with anti-human IL-2R alpha-chain mAb. Solid phase ELISA analysis of serum samples from mice bearing mJB6 lymphoma showed high concentrations of human sIL-2R. None of the control mice without lymphoma or with human nonlymphoid tumors (prostatic carcinoma, ovarian carcinoma, and glioblastoma multiforme) showed detectable human sIL-2R. The sIL-2R serum titers of mJB6-bearing mice correlated strongly with tumor volume (P < 0.0001). Tumors as small as 0.4 to 0.8 mm3 could be detected by this method. The sensitivity of sIL-2R ELISA exceeded at least 150 times the sensitivity of conventional radioisotopic tumor detection. Total resection of mJB6 tumors resulted in complete clearance of sIL-2R from the murine serum within 48 hours with a half-life of 6 hours. Accordingly, partial resection led to a significant decrease in sIL-2R followed by gradual increase with tumor regrowth. sIL-2R was also detected in the urine of mJB6-transplanted mice. As in serum, urine concentrations of sIL-2R were proportional to tumor mass (P < 0.02). Based on these findings we postulate that malignant cells are a major source of serum sIL-2R in patients with lymphoid tumors. In addition, our data further support monitoring sIL-2R concentration in body fluids as a sensitive method to detect change in tumor volume in such patients.

Primary cutaneous Hodgkin's disease. Unique clinical, morphologic, and immunophenotypic findings.

The authors studied five patients with primary cutaneous Hodgkin's disease (PCHD). Each patient presented with skin lesions without evidence of systemic HD. Skin lesions were papules or nodules, many of which regressed spontaneously. Lesions were distinguished from lymphomatoid papulosis (LyP) by the presence of numerous diagnostic Reed-Sternberg (RS) cells that expressed CD30 and CD15 but were negative for CD45R; LyP lesions usually are CD15-, CD45R+. Anaplastic large cell lymphoma (ALCL) was excluded by the polymorphous background of inflammatory cells in PCHD. Three patients with PCHD had a benign course without systemic disease with up to 20 years of follow-up, whereas two other patients developed mixed-cellularity HD in lymph nodes 2 months and 6 years following the onset of PCHD. This study indicates that PCHD does occur as a rare but distinct clinicopathologic entity morphologically and immunophenotypically indistinguishable from nodal HD but with an unexpectedly indolent course in some patients. Patients with PCHD should be observed for development of systemic HD, but unlike patients with LyP or ALCL, an association of PCHD with mycosis fungoides or cutaneous T-cell lymphoma has not yet been observed.

A small-cell-predominant variant of primary Ki-1 (CD30)+ T-cell lymphoma.

We describe nine patients with a primary Ki-1 (CD30)+ T-cell lymphoma containing numerous, often CD30-negative, small lymphocytes with irregular nuclei and a minor population of large CD30+ tumor cells. All previously described primary Ki-1+ lymphomas have been large-cell neoplasms. In this small-cell variant, the diagnosis of lymphoma was difficult to make because there was a predominance of small lymphocytes and, in some cases, clinical features suggested an inflammatory process. Patients were young (age range 0.3-40 years, median 14 years), and frequently had B symptoms (56%); sites of involvement were predominantly skin (78%) and lymph node (67%). The actuarial 2-year disease-free survival was 14%, and the overall survival was 51%. Two patients had a rapidly fatal course. In all cases histologic sections showed a predominance of small lymphocytes with marked nuclear irregularity and often a perivascular/intravascular distribution of CD30+ large cells. All cases had a T-cell phenotype. In four cases the large and small cells could be compared and had a similar aberrant T-cell phenotype. Large cells were CD30+, but only rare small cells expressed CD30. Cytogenetic studies revealed a t(2;5)(p23;q35) in four of four cases studied. Four patients had numerous large cells on repeat biopsies; two of these developed sheets of large CD30+ cells typical of anaplastic large-cell lymphoma (ALCL). These cases provide further evidence that primary Ki-1+ lymphoma has a morphologic spectrum that includes a small-cell variant. Although very different morphologically from previously described Ki-1+ ALCL, this small-cell variant is clearly part of the disease spectrum on the basis of clinical features, the presence of the t(2;5)(p23;q35), the aberrant T-cell phenotype in the small and large cells, as well as histologic progression seen in several patients.

Perivascular Ki-1 + lesions.

We describe three male patients diagnosed by histologic and immunophenotypic criteria to have a Ki-1 + lymphoproliferative disorder. All three cases shared a unique morphologic finding, not previously described: prominent perivascular cuffing of anaplastic/pleomorphic tumor cells around small and medium sized vessels. One case was a Ki-1 + anaplastic large cell lymphoma (ALCL) which developed in the setting of a low grade follicular B-cell lymphoma. A pseudoglandular pattern caused referral for consultation as a possible adenocarcinoma. One case was a cutaneous Ki-1 + lymphoproliferative disorder consistent with lymphomatoid papulosis (LyP). The third case was a cutaneous Ki-1 + ALCL. This study provides evidence that although perivascular cuffing of tumor cells is not frequently seen in Ki-1 + lymphoproliferative disorders (3 of 116 cases in our consultation file), it may be a prominent architectural feature causing confusion with epithelial or mesenchymal tumors. For this reason we recommend inclusion of this feature in the list of architectural features already described for Ki-1 + lymphoproliferative disorders.

Primary hepatic carcinoid tumor. An electron microscopic and immunohistochemical study.

A case of primary carcinoid tumor of the liver with striking morphologic and electron microscopic features is reported. Conventional histologic examination showed a prominent paranuclear clear zone in numerous tumor cells. By electron microscopic examination, this clear zone corresponded to a paranuclear mass of intermediate filaments admixed with neurosecretory granules and other cytoplasmic organelles.