Thyroid carcinomas with a variable insular component

Prognostic significance of histopathologic patterns

Authors


Abstract

BACKGROUND.

An insular growth pattern may be observed focally both in papillary and follicular thyroid carcinoma. The aim of the current study was to determine whether a greater extension of the insular component (IC) influences different clinical and histologic features at diagnosis, and a different tumor aggressiveness in terms of frequency in the occurrence of metastases as well as survival.

METHODS.

Thirty-three patients with histopathologic findings consistent with IC were included in the study. IC was focal (<50% of the tumor area) in 16 patients and predominant (>50% of the tumor area) in 17 patients. These 2 groups were compared with a control group of 66 patients with differentiated thyroid carcinoma.

RESULTS.

At diagnosis, carcinomas with predominant IC differed from those with focal IC with regard to greater tumor size and a higher frequency of extrathyroidal extension and distant metastases. Patient follow-up ranged from 5 to 188 months. The cumulative rate of distant metastases was significantly higher in patients with predominant IC. At the time of last follow-up, carcinomas with predominant IC demonstrated a lesser frequency of disease-free outcome (P = .002) and a higher number of tumor-related deaths (P = .002), either when distant metastases were present (P = .03) or absent (P = .05) at the time of diagnosis.

CONCLUSIONS.

The presence of predominant IC is associated with a poor prognosis in terms of ongoing disease or death. Predominant IC should be considered a separate entity from not only the classical papillary or follicular carcinomas but also the focal IC tumor. Cancer 2007. © 2007 American Cancer Society.

Insular thyroid carcinoma was first described in 1984 by Carcangiu et al.,1 who defined its histologic diagnostic criteria as referring to a characteristic appearance consisting of well-defined nests or “insulae” of tumor cells with round “dark” and monomorphic nuclei and scant cytoplasm. This tumor is relatively infrequent, accounting for only 3.8% to 6.2% of all thyroid carcinomas in most series1–3 and is generally considered to have a poor prognosis, with a 10-year mortality rate ranging from 9% to 61%.3–6 For this reason, a more aggressive therapy is generally recommended. An insular growth pattern may be observed focally in both papillary and follicular carcinomas.7 However, the prognostic and therapeutic significance of the focal insular component (IC) versus the predominant one remains unclear. In the study by Ashfaq et al.3 that aimed to clarify the significance of an insular pattern within follicular or papillary carcinoma, the authors concluded that IC alone does not appear to affect the tumor's biologic behavior. Other authors emphasize that the insular growth pattern within a well-differentiated thyroid carcinoma (DTC) correlates with poor prognosis regardless of its extent.8–10

The aim of the current study was to investigate whether a greater extension of IC, which allows the distinction between focal and predominant, influences: 1) different clinical and histologic features that can be observed at the time of diagnosis, and 2) a different tumor aggressiveness, evaluated by means of a sufficiently long follow-up and the comparison with a control group in terms of survival as well as frequency of the occurrence of recurrent tumor or metastases.

MATERIALS AND METHODS

Among 764 patients with nonmedullary thyroid cancer who underwent radioiodine therapy after thyroidectomy at Policlinico Gemelli Hospital in Rome, Italy, from 1992 to 1998, 33 patients had histopathologic findings demonstrating an IC according to the criteria of Carcangiu et al.1

Among the patients who underwent surgery and postoperative administration of 131I for the treatment of DTC during the same period, 66 patients were individually selected and used as a control group. These 66 patients (2 controls for each patient demonstrating an IC) could be matched to the ‘insular’ patients according to age, tumor size, and differentiated component of the tumor. The age and tumor size of patients demonstrating an IC and of the control group are shown in Table 1.

Table 1. Age and Tumor Size of Patients With IC and of the Control Group
 Mean ± SDMedian
  1. IC indicates insular component; SD, standard deviation.

Size, cm
 Patients with IC4.35 ± 1.534.20
 Control group4.33 ± 1.354.25
Age, y
 Patients with IC55.30 ± 1862.0
 Control group56.60 ± 1658.0

Histopathology and Staging

All surgical specimens were submitted to histopathologic evaluation and stained with hematoxylin and eosin. If required, immunostainings for thyroglobulin (Tg), calcitonin, and carcinoembryonic antigen were performed. All slides were reviewed by a pathologist (G.F.) and the most important morphologic prognosticators were reported, in particular the vascular, capsular, and parenchymal invasion. The amount of the IC for each case was assessed by examining 4 representative slides for each case and graded as focal or predominant depending on the percentage of the slide area occupied by the IC. The cutoff for the distinction between the 2 degrees was established at 50%. The differentiated component was classified as follicular or papillary carcinoma. All cases were staged according to the pTNM system,11 in which T (extent of the primary tumor) and N (regional lymph node metastases) were assessed on the basis of their pathology reports, whereas M (distant metastases) was defined on the findings of the first post-131I therapy whole-body scan (WBS) and/or imaging modalities.

Postoperative Management and Follow-up

Patients were treated according to a homogeneous protocol, which included total thyroidectomy and excision of the clinically involved lymph nodes, 131I postoperative administration for thyroid remnant ablation, and L-thyroxine (L-T4) suppressive therapy. Additional 131I therapies were given to treat local recurrences or distant metastases that were not removable by surgery. Criteria for disease remission were negative 131I WBS and Tg < 2 ng/mL after L-T4 withdrawal or recombinant thyroid-stimulating hormone injection. The diagnosis of recurrence was made after observing an elevation of Tg levels associated with focal areas of 131I uptake at WBS and/or evidence of lesions at ultrasound (US) or computed tomography (CT) and/or positive cytology examination. Disease progression was defined as the enlargement of tumor masses, as evaluated by 131I-WBS and/or US and/or CT and/or the appearance of new metastatic sites. Follow-up was obtained in all patients by reviewing the clinical records.

Statistical Analysis

For the statistical comparison between focal and predominant IC at initial diagnosis, the following parameters were considered: sex, age, preexisting multinodular goiter, size of the primary tumor, lymph node metastases (at pathologic examination), vascular invasion, distant metastases, and Tg value. The Student t test, chi-square test (χ2), Fisher exact 2-tailed test, and the Mann-Whitney U test were used. For statistical analysis, age >45 years (as proposed in the 6th edition of the TNM classification system) and a tumor size >4 cm (representing the limit between pathologic T2 and T3, according to the American Joint Committee on Cancer Staging manual) were considered the cutoff values.11, 12 Logistic regression (backward stepwise regression) analysis was used to compare the distribution by considering predefined parameters and disease-specific death. The following 8 parameters were tested: age (≤45 years or >45 years), size of the primary tumor (≤4 cm or >4 cm), multifocality, vascular invasion, extrathyroidal extension, lymph node involvement at presentation, bone metastases at presentation, and tumor histotype (focal or predominant IC or DTC). For the evaluation of follow-up, the cumulative survival curves were plotted by the Kaplan-Meyer method. Differences between curves were tested using the log-rank test (Mantel-Cox) and Breslow test (generalized Wilcoxon). The significance was set at the 5% level (α = 0.05). Data were analyzed with the SPSS 13.0 statistical software for Windows (SPSS Inc., Chicago, Ill). This retrospective study was approved by the local Ethics Committee.

RESULTS

The prevalence of cases with IC was 4.3%. Of the 33 cases included in this study, 16 patients had a focal IC (10 papillary or follicular variant of papillary carcinomas and 6 follicular carcinomas) and 17 patients had a predominant IC (7 papillary or follicular variant of papillary carcinomas and 10 follicular carcinomas) (Figs. 1 and 2a–c). Among the predominant IC carcinomas, the percentage of IC was ≥70% in all specimens. A summary of the results regarding the comparison of clinical and histopathologic findings between patients with focal IC or predominant IC is shown in Table 2. Carcinomas with predominant IC differed at diagnosis from those with focal IC in terms of statistical significance, greater tumor size, and a higher frequency of extrathyroidal extension and distant metastases, as well as for a higher Tg level.

Figure 1.

A focal insular pattern within a follicular variant of a papillary carcinoma; the insular component can be seen in the top right corner of the photograph (H&E, ×250).

Figure 2.

(a) Low magnification of a predominant insular carcinoma surrounded by a thick fibrous capsule (H&E, ×50). (b) The same case at medium power with the distinctive solid pattern (H&E, ×125). (c) High-power details of the tumor structure comprised of ‘insulae’ of monomorphic cells with dark nuclei. The arrow points to a mitotic figure (H&E, ×250).

Table 2. Clinical and Histopathologic Findings in Patients With Focal IC and Predominant IC
 Focal ICPredominant ICP
  • IC indicates insular component.

  • *

    Determined using the Fisher exact 2-tailed test.

  • Determined using the Mann-Whitney U test.

  • Determined using the Student t-test.

Sex6 men 10 women7 men 10 womenNS*
Age, y61 ± 8.563 ± 9NS
Previous thyroid disease9/167/17NS*
Size of primary tumor, cm3.6 ± 1.55 ± 1.2.006
Multifocality4/162/17NS*
Vascular invasion10/169/17NS*
Extrathyroidal extension1/167/17.039*
Lymph node involvement2/164/17NS*
Distant metastases0/165/17.04*
Thyroglobulin, ng/mL9 ± 5.2292 ± 289.9.008

Diagnostic and Therapeutic Data

After surgery, all patients were treated with 131I. Table3 reports the occurrence of locoregional and distant metastases in the patients in these 3 groups. Lung metastases were more frequent in patients with predominant IC; the difference, however, was not statistically significant. Bone metastases were more frequent among patients in the predominant IC group compared with the control group (P = .048); no patient with focal IC developed bone metastases. Distant metastases at unusual sites occurred in 1 patient in the control group who developed a single liver metastasis after 35 months from the time of the initial diagnosis, and in 1 patient with predominant IC who had bone marrow infiltration at presentation and demonstrated rapid progressive disease with brain and liver involvement; the latter case has been described previously.13

Table 3. Occurrence of Locoregional and Distant Metastases in Patients With Focal IC and Predominant IC and Patients in the Control Group
 Focal ICPredominant ICControl groupP
  1. IC indicates insular component; LN, lymph nodes; NS, not significant.

LN metastases
 At diagnosis2 (12.5%)4 (23.5%)24 (36%)NS
 During follow-up2 (12.5%)4 (23.5%)4 (6 %).051 (predominant vs controls)
 Cumulative4 (25%)8 (47%)28 (42%)NS
Distant metastases
 At diagnosis0 (0%)6 (35%)10 (15%).018 (predominant vs focal)
 During follow-up2 (12.5%)3 (18)4 (6%)NS
 Cumulative2 (12.5%)9 (53%)14 (21%).036 (predominant vs focal)
   .011 (predominant vs controls)

Radioiodine uptake in neoplastic lesions was present in 3 of 5 cases with focal IC, in 9 of 15 cases with predominant IC, and in 21 of 29 cases in the control group (P value was not significant). An 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) scan was performed after L-T4 withdrawal in 4 cases with a negative 131I WBS (1 patient in the control group, 1 patient with focal IC, and 2 patients with predominant IC), and was positive in all cases.

Additional 131I therapies were used to treat local recurrences or distant metastases that were present at diagnosis or occurred during follow-up. No patient with distant metastases was cured with 131I therapy. However, 4 of 9 patients (44%) in the predominant IC group, 1 of 2 patients in the focal IC group, and 11 of 14 patients (79%) in the control group achieved some benefit from radioiodine therapy, with disease stabilization or a reduction of tumor lesions. Additional therapy (surgery and/or radiotherapy and/or chemotherapy) was performed in 3 patients with focal IC, 6 patients with predominant IC, and 22 patients in the control group.

Follow-up and Outcome

At the time of last follow-up, patients with predominant IC demonstrated a lesser frequency of disease-free outcome (P = .002) and a higher number of tumor-related deaths (P = .002), regardless of whether distant metastases were present (P = .03) or absent (P = .05) at the time of diagnosis (Tables 4 and 5). One patient in the focal IC group (6%) and 4 patients in the control group (6%) died of causes unrelated to the disease.

Table 4. Follow-up and Outcome of Patients With IC (Either Focal IC or Predominant IC) and Patients in the Control Group
 Focal ICPredominant ICControl groupP
  1. IC indicates insular component; NS, not significant.

Follow-up, mo
 Range55–1425–16315–188 
 Mean108 ± 2187 ± 52100 ± 45 
 Median106103100.4 
Disease free12 (75%)4 (24%)45 (68%).002
Alive with disease2 (12.5%)6 (35%)11 (17%)NS
Dead of disease1 (6.25%)7 (41%)6 (9%).002
Dead of unrelated causes1 (6.25%)0 (0%)4 (6%)NS
Table 5. Follow-up and Outcome of Patients With Insular Carcinoma (Either Focal IC or Predominant IC) and Patients in the Control Group According to the Presence or Absence of Distant Metastases at Diagnosis
 Focal ICPredominant ICControl groupP
  1. IC indicates insular component; NS, not significant.

Distant metastases at diagnosis0/16 (0%)6/17 (35%)10/66 (15%) 
Mean survival58 ± 61 mo101 ± 52 mo 
Disease free0 (0%)2 (20%)NS
Alive with disease2 (33%)5 (50%)NS
Dead of disease4 (67%)3 (30%).03
No distant metastases at diagnosis16/1611/1756/66 
Mean survival108 ± 21 mo104 ± 45 mo100 ± 31 mo0.24
Disease free12 (75%)4 (36%)43 (77%)NS
Alive with disease2 (12.5%)4 (36%)6 (11%).054
Dead of disease1 (6.25%)3 (28%)3 (5.5%)NS
Dead of unrelated causes1 (6.25%)0 (0%)4 (6.5%) 

Logistic Regression and Survival Curves

On logistic regression analysis, 3 variables were found to be statistically correlated with death: male sex, distant metastases at the time of diagnosis, and predominant IC (Table 6). Figure 3a shows the survival curves for the focal IC, predominant IC, and control groups. The survival of the focal IC patients was quite similar to that of the controls (survival rate at 10 years of 91% vs 88%), whereas survival was remarkably reduced in the patients with predominant IC (59%). The difference was statistically very significant (P < .001 by the log-rank test and Breslow test). Among patients with no distant metastases at the time of diagnosis, the survival of the focal IC group was found to be identical to that of the controls (90% at 10 years), whereas survival of the predominant IC group was reduced to 72% (statistical difference: P = .077, log-rank test; P = .01, Breslow test) (Fig. 3b). Among patients with distant metastases at the time of diagnosis, the survival of the predominant IC group was found to be reduced (33% at 10 years vs 70% of the control group [P = .074, log-rank test; P = .055, Breslow test; the difference was only of borderline statistical significance]) (Fig. 3c).

Figure 3.

Overall survival (Kaplan-Meier plots) in patients with focal insular component (IC) and predominant IC and in patients from the control group. (a) Survival in all patients from the 3 groups (P < .001, log-rank test and Breslow test). (b) Patients without distant metastases at the time of diagnosis (P = .077, log-rank test; P = .01, Breslow test). (c) Patients with distant metastases at the time of diagnosis (P = .074, log-rank test; P = .055, Breslow test).

Table 6. Results of Logistic Regression Analysis
VariableCoefficientPOR95% CI
  1. OR indicates odds ratio; 95% CI, 95% confidence interval; IC, insular component.

Male sex1.96.0167.161.43–35.79
Distant metastases at diagnosis2.26<.019.631.93–48.10
Predominant IC1.85.0186.411.37–30.30

DISCUSSION

The current series refers to patients with IC who were diagnosed according to the morphologic criteria of Carcangiu et al.,1 which was represented as either a predominant pattern or a minor component. Different arbitrary definitions regarding the amount of the IC are given in the literature. Ashfaq et al.3 defined tumors with a minor IC when the IC constitutes 10% to 40% of the tumor area and predominant IC when IC is present in 50% to 75% of the tumor area. Similar percentages were used by Papotti et al.4 Nishida et al.9 analyzed a broad spectrum of poorly differentiated thyroid carcinomas, dividing them into focal when the poorly differentiated component accounted for < 10%, and diffuse when that component accounted for ≥10% of the tumor.

In the current series, an arbitrary cutoff value of 50% of the IC component was chosen, even though among predominant IC carcinomas the percentage of the IC was ≥70% in all cases. The cutoff value of 50% of the IC component was regarded as the best possible compromise between data from the literature4, 10 and the ability to evaluate the amount of tumor with a semiquantitative method based on a simple histologic observation.

The literature reports a nearly general consensus regarding the clinical behavior of tumors with a pure or predominant IC (properly insular carcinomas). Although all publications refer to a small number of patients, the conclusion most frequently reported is that thyroid carcinomas with a predominant insular pattern demonstrate a more aggressive clinical course, with a higher incidence of regional invasion and distant metastases, as well as a higher mortality than well-differentiated papillary or follicular carcinomas.6, 14–17 When examining thyroid carcinomas with a variable IC, there are discrepancies in the biologic aggressiveness and prognostic relevance of this histotype, mainly related to its rarity. Ashfaq et al.3 did not find any correlation between the entity of IC and the stage of disease and patient outcome in terms of regional or distant spread. Although those authors observed a trend toward higher mortality in patients with predominant IC, this phenomenon was far from being statistically significant. Conversely, Papotti et al.4 demonstrated that a predominant insular pattern significantly increases the risk of disease recurrence compared with those patients demonstrating a minimal IC. Other authors report that tumors showing an insular growth pattern have a more aggressive clinical course when compared with DTC, irrespective of the extent of the IC.8–10 Finally, according to Albores-Saavedra et al.18 a minor IC that comprises <5% of the tumor mass is an additional feature of the macrofollicular variant of papillary thyroid carcinoma, and does not appear to adversely affect the excellent prognosis of this tumor.

The results of the current study indicate that a predominant IC pattern in either papillary or follicular thyroid tumors demonstrates clinical and histologic features at the time of presentation that differ from those of the focal IC pattern and has a worse patient outcome. According to clinical and pathologic data that are observed at initial diagnosis, the main risk factors that negatively influence the prognosis of DTC include older age, large tumor size, extrathyroidal extension, vascular invasion, and the presence of distant metastases.19, 20 Analyzing these features, both focal and predominant IC were more frequent in patients age >45 years, with no difference noted between the 2 groups. This is in agreement with data from the literature that report a higher frequency of insular thyroid tumors in older patients, although cases of patients age <18 years are occasionally reported.21–26 These cases appear to demonstrate a clinical outcome that is poorer than that observed in DTC diagnosed in childhood. The current series included 2 adolescent patients (a 14-year-old boy and a 15-year-old girl), both of whom were in the focal IC group and had a favorable outcome.

We found that predominant IC was associated with a more advanced tumor stage at the time of presentation that was regarded as a primary lesion measuring >4 cm, and had a higher frequency of extrathyroidal extension and distant metastases. Considering these data, it could be suggested, in accordance with other authors,3, 27, 28 that the aggressive behavior of insular carcinomas may be related to a higher disease stage at the time of presentation rather than to the insular growth pattern. To determine whether the outcome of insular carcinoma was dependent on prognostic factors such as older age and large tumor size, we compared the prognosis of patients with focal IC and those with predominant IC with the prognosis of a group of patients with DTC who could be matched according to age, tumor size, and differentiated component of the tumor. We observed a significantly poorer long-term outcome in patients in the predominant IC group in terms of the cumulative rate of distant metastases and disease-free patients and tumor-related deaths.

With respect to other series,2, 16, 28 we surprisingly observed a relatively lower occurrence of cumulative lymph node involvement, both in the focal IC and the predominant IC group. However, in patients with predominant IC, the cumulative rate of distant metastases was 53%, which is in keeping with previously reported data and is significantly higher when compared with patients in the focal IC group (12.5%) and the control group (21%).

In agreement with the findings of Pellegritti et al.,6 our data also appear to confirm that the adverse prognosis of patients with predominant IC is related to the insular growth pattern per se, rather than to the higher stage of disease at the time of presentation. In the predominant IC group, the rate of disease-free patients was significantly lower and the rate of tumor-related deaths was significantly higher (24% and 41%, respectively) when compared with the focal IC group (75% and 6.25%, respectively) and the control group (68% and 9%, respectively) (P = .002). When patients were stratified for the presence of distant metastases at the time of initial diagnosis, a poorer survival was still associated with the predominant insular growth pattern.

An insular growth pattern is considered in most series as poorly differentiated because of its architecture, which is characterized by the lack of papillae or follicles. However, evidence of follicular differentiation is frequently reported in terms of Tg production and 131I uptake.1, 4 We found a heterogeneous pattern of 131I uptake and a heterogeneous response to therapy. Uptake of 131I in metastatic lesions was present in 60% of patients in the predominant IC group, with a behavior not dissimilar to those in the control group and it was higher in patients who were treated shortly after surgery compared with patients who were treated for recurrent metastatic disease.

The maintenance of 131I uptake in metastatic sites of some patients with the insular histotype allowed treatment with radioiodine. No patient with distant metastases was cured using 131I therapy. However, 44% of patients in the predominant IC group achieved some benefit from radioiodine therapy, compared with 79% of the control group. Because the majority of patients were treated with a homogeneous therapeutic approach, the lower therapeutic response of patients with predominant IC could be correlated with a lower frequency of 131I uptake in metastatic lesions. However, the rate of patients demonstrating 131I uptake in metastatic sites was similar in the 3 groups. It is likely that clinical characteristics or other factors such as timing of 131I administration or 131I kinetics in metastatic lesions could be involved in determining the therapeutic response.29

According to the recent observations of Wang et al.30 and Robbins et al.,31 additional prognostic information for patients with metastatic thyroid carcinomas could be supplied by FDG-PET scan results, and these data also might be applied to insular carcinomas. In fact, the high metabolic activity disclosed by FDG uptake reflects the presence of poorly differentiated tumor cells that are relatively resistant to high activities of 131I, and suggest that other therapeutic options in addition to or in place of radioiodine should be considered.

To our knowledge, the role of molecular mechanisms and genetic events in determining the growth and biologic aggressiveness of insular thyroid carcinoma remains to be explained. Pilotti et al.2 reported somatic point mutations of the ras gene family in cases of insular carcinomas (CAA → AAA transversion at codon 61); however, the same abnormality has been found with similar frequency in the widely invasive variant of follicular thyroid carcinoma.32 Likewise, mutations of the p53 gene are also not specific, because they have been described both in patients with IC and in patients with anaplastic carcinoma.33, 34

In conclusion, data from the current study demonstrate that a greater extension of the IC does affect the biologic behavior per se of DTC and indicates an unfavorable long-term prognostic outlook, thus suggesting the prognostic utility of IC quantification in the histologic report. From a practical point of view, we suggest that predominant IC must be considered a separate entity with respect to classical papillary and follicular carcinomas and the focal IC tumor, and requires a more aggressive therapeutic approach. Conversely, the presence of a focal IC does not appear to affect the excellent prognosis of patients with thyroid tumors.

Ancillary