Nonlinear Diagnosis of Thyroid Gland Pathology

K.M. Beznogov,

L.V. Kundratyev,

S.N. Pauli

Introduction

The morphological diagnosis of thyroid gland pathology during surgery is rightfully

considered one of the most important and complicated tasks faced by anatomist and

surgeon. This research aimed to choose an optimum surgical tactics with tumors and

tumor-like lesions of the thyroid gland, which is achieved by accurate verification of the

process as well as by determining its spread in the organ and/or beyond it. The

experience in the use of nonlinear computer diagnosis (NLS) in surgical clinics covers

quite a short period of time, during which quite conflicting opinions with respect to its

efficiency were formed. Among the prime considerations against an extensive use of

NLS we should mention the possibility to preclude in some instances malignant pattern of

a new growth because of a morphological similarity of follicular tumors. Without down-

grading this problem and based on our own experience in the use of NLS we have

attempted to access the importance of this method for choosing the optimum surgical

tactics and working out some methodical techniques enhancing efficiency and accuracy

of the NLS investigations.

Subject and investigation methods

We analyzed the results of 682 patho-histological investigations of the thyroid gland

carried out in 2000-2001 in patients operated on for solitary nodes, diffuse and

multinodal hyperplasia and autoimmune thyroid diseases. 326 of all surgeries were

accompanied by NLS. NLS data were compared with the final results of

pathohistological investigations.

Result analyses

Of 682 surgical operations of the thyroid gland, 326 (47.8%) were accompanied by

NLSinvestigation. According to our information, there is an increased demand for NLS,

which has to do with a growth of surgical operations for nodular goiter from 70% to 85%

and also with a growing thyroid cancer incidence including hyper-plastic and

autoimmune lesions of the thyroid gland.

We did not succeed in specification of the pattern of the process in the course of 3.6% of

all the surgical operations. In 65% of cases for which diagnosis was postponed, a

malignant process was detected in the final phase of the investigation. An erroneous

intra-operative diagnosis was made in 4.8% of cases with hyper-diagnosis of thyroid

cancer recorded in 5 cases. Carcinoma was not identified during 38 operations, in 23

cases of the surgery the tumor did not exceed 2.5 cm nor did it spread beyond the thyroid

gland. 17 (2.3%) patients needed correction of the amount of the thyroid gland resection,

51which was done on the 4-5th day after the first surgery. The NLS sensitivity was 76.4%,

specificity - 87.6% and accuracy - 78.6%.

The submitted data generalizing the experience in the use of NLS in a specialized

surgical clinic are indicative of extensive opportunities for choosing the optimum surgical

tactics to treat the goiter using this method, and also of a growing recognition of NLS

despite some recent publications questioning the efficiency of consultations during the

surgery. Apart from some evidence of a high efficiency rate, another advantage of NLS

method is indisputably the low percentage of delayed diagnoses, which in our cases was

under 1.4%. The advantages of the method include quickness (10 or 15 min) and relative

technical simplicity of the investigation.

According to our research and some literary evidence, the problems in the course of

NLSinvestigation are caused by differential diagnosis between cellular follicular adenomas

and minimum invasion follicular cancer. These very cases account for the major share of

delayed and erroneous results. In the series of our investigations in 48 cases of tumors with

a micro follicular or trabecular structure (among which 21 were benign and 27

malignant) drawing a final conclusion on their pattern was impossible. In 5 cases it was false

negative. At the same time 257 follicular adenomas and 30 cases of follicular

cancer were correctly verified in the course of surgery.

It is known that the problems of differential diagnosis of follicular adenoma and follicular

cancer are closely related to histo-typical and cylo-typical similarity of the two processes,

that are so much expressed that it is impossible to diagnose a carcinoma without apparent

manifestations of a malignant potential in the form of the tumor germinated capsule with

the tumor invasion into its vessels. With cancer having minimum invasion the nidi of

infiltrative growth appear to be isolated. In addition, being microscopically invisible

these diagnostically important areas may escape observation in case of a limited number

and random choice of investigation targets in the course of surgery.

We have considered a number of clinical and microscopic characteristics in terms of their

potential use in NLS differential diagnosis of follicular tumors of the thyroid gland. A

comparison was made considering age and gender of patients as well as the size of the

tumor nodes in 61 cases of follicular cancer and 162 cases of follicular adenoma. This

study of the parameters didn’t detect any difference between these two patient groups.

The male/female ratio in both groups was the same -1:9, the average age of the patients

operated on for adenoma was 42.36+13.76 and was not different from that of the patients

in the follicular cancer group (41.40+16.14). Follicular cancer is known to be more

common with elderly people and very rare with children and teenagers. The latter

circumstance could be a supplementary reference point for investigating solitary nodes of the

thyroid gland in junior patients. The investigation analysis of 89 cases of follicular tumors in

patients operated on at the age of 30 showed that in one third of cases the new growth was

of a malignant nature.

Some differences were found in the average diameter of the tumor nodes: 3.05+1.45 cm

for adenoma and 3.89+1.77 for follicular cancer (p <0.05). At the same time, the

52coincident size limits (from 1.5 to 8 cm) in patients in both groups made this evidence an

unreliable indication in the differential diagnosis of tumors. Both kinds of neoplasms

equally often (approx. in 80% of cases) were not accompanied by morphologically

significant changes in the thyroid gland being a solitary node. Some frank secondary

changes such as sclerosis, petrifaction, cystic changes, hemorrhages, etc., were more

often observed in follicular adenomas, however these distinctions were not authentic

enough.

In our opinion the difficulties in the clinical morphological interpretation of the follicular

tumors pattern need more than anything else improvement in the methodical techniques,

which is especially important considering certain time and hardware-related limitations of

the NLS-method. When faced with diagnostic difficulties we investigate series of nidi in

every 30-60 mm. Conducting investigation on such a large scale consumes additional

time (30-45 min) and yet in many instances it enables to specify the pattern of a follicular

tumor. If the investigation of some additional nidi dosen’t produce the desirable result,

the diagnosis is performed after the surgery. According to our observations, in 65%

delayed cases the tumor proved to be malignant yet had minimal manifestations of

invasive growth into the node capsule or its individual vessels. Absence of palindromium

in 95-99% of cases, following the surgery for follicular cancer with minimal

manifestations of invasive growth into the capsule and with some individual vessels (up

to 5) involved, gives solid grounds to classify these tumors as clinical “boundary”

processes whose malignant potential remains conditional and justifies the tissue-sparing

amount of the thyroid resection (lobectomy with isthmectomy and subtotal

thyroidectomy) similar to the one recommended for surgery for follicular adenoma.

Thus, the NLS-investigation of follicular thyroid tumors can be regarded as an efficient

method for choosing the optimum surgical approach, because with sufficient experience

and proper performance it enables the diagnosis of some clinically adverse forms of

follicular thyroid cancer that actually need radical surgery and post surgical treatment.

Papillary cancer is the most common form of thyroid gland carcinoma. It was correctly

diagnosed by means of the NLS-method in 63.2% of cases, was a diagnosis failure in

26.3% and was responsible for delayed diagnosis in 0.6% of the observations. Unlike

follicular tumors, most versions of papillary cancer typically have frank histo-typical

differences from benign proliferate processes, clear manifestations of infiltrative

germination in the tumor-surrounding tissues and frequent metastatic lesion of the lymph

nodes by the time of surgery, which enables the diagnosis of a malignant process without

difficulty even with inadequate practical experience. The difficulties we confronted

mostly concerned papillary micro carcinoma, which made 71% of cases non diagnosed

during the surgery, and also a follicular encapsulated version of papillary cancer that was

responsible for the rest 29% of diagnostic errors.

The problem of NLS papillary micro carcinoma lies in a macroscopic search for a cancer

nidus in the removed fragment of the thyroid gland, which creates difficulties because of

concomitant changes in the thyroid gland at a multi-nodal form of the goiter or

autoimmune processes. At the same time in 45 cases micro-carcina was diagnosed intra-

53operatively, including 15 cases, where the tumor size was below 0.4 cm. In 23 cases of

papillary micro carcinoma unidentified by NLS, the surgery was performed for

multinodal goiter (15 cases) and autoimmune thyroiditis (8 cases), which determined the

required amount of surgical intervention (subtotal or total thyroidectomy). The

undiagnosed micro cancer nidi sized from 0.4 to 0.9 cm did not spread beyond the thyroid

gland and afterwards none of the patients affected by latent carcinoma required another

surgical operation to extend the amount of thyroid gland resection.

Papillary micro carcinoma is known to localize quite often in the thyroid gland,

especially in elderly people and it does not always display its malignant potential in the

form of clinical implications (I.L. Avetisyan, 1999). The progress of the great majority

of such tumors is entirely favorable. Meanwhile, a direct relationship was established

between the size of a papillary micro carcinoma nidus and the frequency of its

metastasizing into the cervical lymph nodes. According to our information and some

evidence obtained from other clinics, a tumor in excess of 0.5 cm incurs a higher risk of

potential palidromium and requires a more radical approach to its treatment, than a

microscopic cancer nidus. In this connection, it is recommended to take a series of

parallel shots of the thyroid tissue in different projects in the course of macroscopic

search for cancerous nidi. According to our observations this technique may be effective

in diagnosing 52% of papillary micro carcinomas sized up to 0.4 cm and 68.6% of tumors

over 0.4 cm, and in most cases will ensure the right choice of a surgical approach.

According to most pathologists, diagnosis of follicular version of papillary cancer is

considered to be one of the most difficult problems in NLS-investigations. Erroneous

verification of this neoplasm often leads to another surgical operation. Among the

difficulties in diagnosing this kind of tumor, we should first mention some artificial

histogram changes in tumor cells, which hamper identifying diagnostically significant

cytological criteria of papillary cancer. In our series of observations follicular version of

papillary cancer was reported in 43 cases, in 2 of which diagnosis was postponed until

after final investigations and in 5 (3.2%) cases the diagnosis was false negative. None of

the cases showed any metastatic lesions of cervical lymph nodes. Considering the

spectral similarity of follicular encapsulated version of papillary cancer to adenomatous

goiter, the cytological differences in differential diagnosis are the decisive criterion in

differential diagnosis. In order to assess them in doubtful cases we have additionally

investigated some impression smears, which in most cases helped detect some changes in

the nuceli characteristics of papillary thyroid cancer, such as irregular shape, jagged

boundaries, deep nucleolemma invaginations, outlines of intranuclear sulci and inclusions

marked off by marginally condensed chromatin, fine chromatin dispersion, etc., as well

as to indirectly assess some cohesive properties of tumor cells and the inflammatory

infiltrate pattern. Among some cancer-suspicious histological signs observed on a frozen

section, we can note polymorphism of follicles lined with high cubical epithelium with

intensely tinged colloid (if fixed in ethanol!) and/or its marginal vacuolation, close

adherence of follicles to one another owing to scarce stroma in the central part of the

node, hemorrhage in the follicle lumen at abundance of siderophages, multinuclear cells,

etc.

54The recent years publications have extensively debated expediency of NLS-investigations

of the thyroid gland in the cases with the available results of aspiration biopsy. The

cytological investigation is known to be the most extensively applicable method of pre-

surgical diagnosis of nodular forms of the goiter because of its accessibility,

comparatively low cost, lack of traumatism and most importantly, high accuracy. The

progress in the diagnosis of thyroid tumors using the cytological investigation technique

has induced some researchers to become result-oriented in choosing a surgical approach.

In this connection we made an attempt to define to what extent this approach is justified

after having analyzed our own observations and the practical experience of specialists in a

number of well-known clinics of the world.

Among some major problems of fine needle aspiration biopsy (FNAB) performed under the

control of ultrasound scanning (US) we can mention an amount of aspirate inadequate for

diagnosis as well as some situations that do not allow to preclude a malignant process in the

node under the investigation. While in the former instance a repeated procedure may

prove efficient for 30% of the patients, the other problem needs to be solved by a surgical

removal of the tumor with a subsequent histological verification of its pattern. According to

some publications, the number of tumors, which malignant potential cannot be excluded

based on FNAB results, exceeds 11%. In the majority of cases (about 70%) this kind of

diagnostic problems is caused by follicular adenoma.

According to our information of 338 patients who had surgery after preliminary FNAB the

malignant process was not cytologically ruled out in 41 cases. In 26 of these

observations follicular adenoma and in 15 thyroid cancer were verified (2 follicular, 3

medullary and 10 papillary carcinomas). In all cases the NLS allowed to specify the

diagnosis and avoid errors in defining the amount of resection.

The problem of FNAB accuracy is no less pertinent. According to some recent reports,

the sensitivity and specificity of thyroid FNAB has approached 100%. At the same time,

it proves to be difficult to interpret the data due to different approaches to their analysis.

For instance, specialists in some clinics, where the results were highly accurate, consider

only specific cytological conclusions ignoring a category of tumors, which cytological

picture gives grounds only to suspect malignant change. In analyzing the FNAB

accuracy some researchers consider follicular adenoma in the same category as malignant

tumors. Though this kind of approach may be justified in terms of indications for

surgery, by no means it can be justified in terms of a surgical approach to be chosen. The

analysis of some publications shows, that FNAB can ensure the right surgical approach

only in 70-75% of cases. The FNAB efficiency data-evaluated after classifying cancer-

suspicious conclusions as malignant tumors and adenoma as a benign tumor, were as

follows: sensitivity - 92.1%, specificity - 94.4%, accuracy - 93.45. Thus, the FNAB

data could help decide on the proper amount of surgery for 90% of the patients, which in

terms of specialized clinic cannot be a sufficient reason and is an argument in favor of

supplementing FNAB with NLS-investigation.

55From comparison of the FNAB and NLS results it was found that 27 false negative

results of punch biopsy 21 nidi of malignant growth were detected during surgery among

multiple benign goiter nodes. The latter, being prevalent in clinical implication, became

the object of FNAB leaving some latent cancer nidi sized 0.2-1 cm undiagnosed before

the surgery. The insufficiently accurate cytological interpretation of the pattern of a

cystic cavity in the 111G also requires an intra-operative verification. In our series in 2

cases of encapsulated papillary cancer with cystic generation wrongly interpreted as a

benign process during FNAB the diagnosis was rectified by means of NLS.

Another equally important task of intra-operative investigation is leveling possible false

positive conclusions of FNAB. In our series of 497 patients having benign new growths

according to FNAB results, papillary cancer was suspected in 2 cases and a malignant

process was not excluded in 26 cases. The rectification of the process pattern in the

course of surgery facilitated selection of the most efficient surgical approach in all cases.

Conclusions

1. The NLS-investigation of thyroid tumors is an efficient method for choosing

surgical approach in surgery for nodal and diffuse forms of goiter.

2. The NLS performed for the patients following FNAB, considerably enhances the

accuracy of morphological investigations at the preliminary phase, and optimizes

the surgical approach in surgery for the thyroid gland.

3. This is a very important reason in favor of the appropriateness of using NLS-

investigation together with FNAB.